The Enzymes: Control by Phosphorylation, Part A : General Features, Specific Enzymes

The Enzymes VOLUME XVII CONTROL BY PHOSPHORYLATION Part A General Features Specific Enzymes (I) Third Edition This P...

Author: Paul D. Boyer

40 downloads 500 Views 32MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

The Enzymes VOLUME XVII

CONTROL BY PHOSPHORYLATION Part A General Features Specific Enzymes (I) Third Edition

This Page Intentionally Left Blank

THE ENZYMES Edited by Edwin G. Krebs Paul D. Boyer Department of Chemistry and Biochemistry and Molecular Biology Insiitute university of California Los Angeles. California

Howard Hughes Medical Insiiiute and Department of Pharmacology University of Washington Seattle, Washington

Volume XVII CONTROL BY PHOSPHORYLATION Part A General Features Specific Enzymes (I) THIRD EDITION

1986

ACADEMIC PRESS, INC. Harcourt Brace Jovanovich, Publishers

Orlando San Diego New York Austin Boston London Sydney Tokyo Toronto

BY ACADEMIC PRESS. INC. ALL RIGHTS RESERVED. NO PART O F THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS. ELECTRONIC OR MECHANICAL. INCLUDING PHOTOCOPY. RECORDING. OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM. WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

COPYRIGHT 0 1986

ACADEMIC PRESS, INC. Orlando. Florida 32887

United Kingdom Edition published by

ACADEMIC PRESS INC.

(LONDON) 24-28 Oval Road. London NWI 7DX

LTD.

Library of Congress Cataloging in Publication Data (Revised for vol.: 17, pt. A) The Enzymes. Includes bibliographical references. 1. Enzymes-Collected works. 1. Boyer, Paul D., ed. [DNLM: 1. Enzymes. Q U 135 B791el QP601.ES23 574.19’25 75-1 17107 ISBN 0-12-122717-0

PRINTED IN THE UNITED STATES OF AMERICA

86 87 88 89

9 8 7 6 5 4 3 2 1

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

Section 1. Enzymology of Control by Phosphorylation

1. The Enzymology of Control by Phosphorylation EDWING . KREES I. 11. 111. IV.

Historical Aspects of Protein Phosphorylation . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . Protein Phosphorylation-Dephosphorylation Reactions ....................... Classification of Protein Kinases and Phosphoprotein Phosphatases . . . . . . . . . . . . . Protein Phosphorylation and the Regulation of Biological Processes . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 10 15 18

2. Cyclic Cascades and Metabolic Regulation EMILYSHACTER, P. BOON CHOCK,SUEGoo RHEE, AND EARLR. STADTMAN I. Perspectives . . . .................................. . . . . . 11. Features of Cyclic Cascade Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Experimental Verification of the Cyclic Cascade Model ...................... IV. Energy Consumption ....................... V. Covalent Interconversion versus Simple Allosteric .................... VI. Concluding Remarks .................................. References ......................... ................

21 21

33 36 31 31

3. Cyclic Nucleotide-Dependent Protein Kinases STEPHEN J. BEEBEAND JACKIE D. CORBIN I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

........................................................... 111. Characterization and Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Mechanism of Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Purification

V

44 46 49 64

vi

CONTENTS V. Biological Role of Protein Kinases ........................................ References ...........................................................

69 100

.

4 Calmodulin-Dependent Protein Kinases JAMES T . STULL.MARYH . NUNNALLY. AND CAROLYN H . MICHNOFF I . Introduction ..........................................................

I1. Myosin Light Chain Kinases ............................................. 111. Multifunctional Calmodulin-DependentProtein Kinases ....................... References ...........................................................

114 118 142 159

.

5 Protein Kinase C USHIOKIKKAWAAND YASUTOMINISHIZUKA I. I1. I11. IV . V. VI . VII . VIII . IX . X. XI .

Introduction .......................................................... Properties .................................. .................. Biochemical Activation ................................................. Physiological Activation ................................................ Action of Tumor Promoters ..................... ........... Inhibitors .................. ........... Synergistic ............................ Growth Response and Down ................ Target Proteins and Catalytic .............. Relation to Other Receptors .............. Conclusion ................................................. ..... References . ....................................................

167 168 169 171 173 174 175 177 179 181 183 183

.

6 Viral Oncogenes and Tyrosine Phosphorylation TONYHUNTERAND JONATHAN A . COOPER I. I1. I11. IV . V. VI .

Introduction and Historical Perspective .................................... Individual Viral Protein-Tyrosine Kinases and Their Cellular Homologues . . . . . . . Other Transformation-Related Tyrosine Phosphorylation Systems ............... General Properties of Protein-Tyrosine Kinases ............................. Cellular Substrates for Protein-Tyrosine Kinases ............................. Conclusions .......................................................... References ...........................................................

192 193

214 219 228 235 237

.

7 The Insulin Receptor and Tyrosine Phosphorylation MORRISF . WHITEAND C . RONALDKAHN I . Introduction and Scope ................................................. I1. Structure of the Insulin Receptor and Its Relation to Other Tyrosine-Specific Protein Kinases ....................................................... I11. Properties of Tyrosine-Specific Protein Kinases .............................

248 249 264

vii

CONTENTS 1V. Tyrosine Phosphorylation of Insulin Receptors and Other Proteins in the Intact Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Regulation of Hormone Receptors by Multisite Phosphorylation . . . . . . . . . . . . . . . . VI. Evidence That Tyrosine Phosphorylation Is Physiologically Important . . . . . . . . . . . VII. Conclusions .......................................................... References ...........................................................

285 296 299 302 302

8. Phosphoprotein Phosphatases LISAM. BALLOUAND EDMONDH. FISCHER 1. Introduction and Historical Overview: The “PR Enzyme” .................... 11. Classification of Protein Phosphatases ..................... Ill. Phosphatase Type 1 (Phosphorylase Phosphatase) . . . . . . . . . . . . IV. Phosphatase Type 2A ........... . . . . . . . . . . . . . . . . . . . . . . . V. Phosphatase Type 2B (Calcineurin) ..................... VI. Phosphatase Type 2C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Phosphotyrosyl-Protein Phosphatases . . . . ...................

References . . .................................... . . . . .

312

331 339 346 349 355

Section II. Control of Specific Enzymes

9. Glycogen Phosphorylase NEILB. MADSEN ................................... cs of Phosphorylase . . . . . . . 111. Molecular Structure

IV. V. VI. VII.

.........................

Substrate-Directed Control of Phosphorylation and Structural Consequences of Serine-14 Phosphorylation ........................ Functional Results of the Phosphorylation of Serine-14 . . . . . . . . . . . . . . . Concluding Remarks . . .................................... References ......................... .............................

377 384 389 390

10. Phosphorylase Kinase CHERYL A. PICKETT-GIES AND DONALA. WALSH .................................. 1. Introduction . . . . . . ................ 11. Physicochemical Properties . . . . . . . . . . . . 111.

396 398

1v.

409

.................................. Catalytic Properties V. VI . Proteolytic Activation of Phosphorylase Kinase . . . . . . . . . . . . . VII . VIII. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CONTENTS

1 1. Muscle Glycogen Synthase PHILIPCOHEN I. 11. 111. IV. V. VI. VII. VIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structure of Glycogen Synthase from Mammalian Skeletal Muscle . . . . . . . . . . . . . . Glycogen Synthase Kinases in Mammalian Skeletal Muscle . . . . . . . . . . . . . . . . . . . Effect of Phosphorylation on the Activity of Skeletal-Muscle Glycogen Synthase . . Synergism between Glycogen Synthase Kinase-3 and Glycogen Synthase Kinase-5 The Glycogen Synthase Phosphatases in Skeletal Muscle . . . . . . . . . . . . . . . . . . . . . Phosphorylation State of Skeletal-Muscle Glycogen Synthase in Vivo . . . . . . . . . . . Interpretation of in Vivo Phosphorylation Experiments . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . , . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

462 462 464 469 47 1 472 478 486 493

12. Liver Glycogen Synthase

PETERJ. ROACH .... . . . . . . . . . . . . . ... . . . . ... . . . . . . . ... . ... . . . . . . . . . . 111. Converting Enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Introduction

500

11. Liver Glycog

501 507

IV. Phosphorylation of Liver Glycogen Synthase V. Comparative Enzymology of Glycogen Syntha YI. Control of Hepatic Glycogen Synthase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Conclusion References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

521 533 534

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

54 1

. . . . . . . . . . . . . . . .. . . ... . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . . .

602

Subject index

Preface Over the past two decades there has been a remarkable increase in the recognition of the salient importance and the wondrous complexity of the control of enzyme catalysis. The modulation of enzymic and other protein-dependent processes by protein phosphorylation or dephosphorylation has emerged as the most widespread and important control achieved by covalent modification. So much information has emerged that adequate coverage in two volumes (XVII and XVIII) was a challenging task. The editors are gratified that the contributing authors have commendably met this challenge. The first portion of Volumes XVII and XVIII concerns the “machinery” of control by protein phosphorylation and dephosphorylation and includes coverage of the major types of protein kinases and of phosphoprotein phosphatases. The central core of the volumes presents chapters on the control of specific enzymes. This is followed by a substantial final section on the control of biological processes. The selection of authors for various chapters was a rewarding experience, but made somewhat difficult because for most topics there was more than one wellqualified potential author. The quality of the volumes was assured by the welcome acceptance of the invitation to participate by nearly all of the invited authors. The reversible covalent modification of enzymes and of proteins with other functions is now known to occur in all types of cells and in virtually all cellular compartments and organelles. Enzymes as a group constitute those proteins whose function and control are best understood in molecular terms. The treatment of enzymes gains additional importance because their regulation provides prototypic examples to guide investigators studying less well defined and often less abundant proteins. The versatility of protein control by phosphorylation finds expression in ion channels, hormone receptors, protein synthesis, contractile processes, and brain function. Chapters in these areas point the way for future exciting developments. Although the breadth of coverage is in general regarded as satisfying, there are other topics or areas that may have warranted inclusion. These include the ix

X

PREFACE

developing knowledge of the control by phosphorylation of histones of the nucleus and the messenger-independent casein kinases, whose role is not as clear as that of the major protein kinases that respond to regulatory agents. The quality of the volumes has been crucially dependent on the editorial assistance of Lyda Boyer and the fine cooperation provided by the staff of Academic Press. We record our thanks here. As readers of this Preface have likely discerned, it is a pleasure for the editors to have volumes of high quality to present to the profession. Paul D. Boyer Edwin G. Krebs

Section I

Enzymology of Control by Phosphorylation


The Enzymology of Control by Phosphorylation EDWIN G . KREBS Howard Hughes Medical Institute and Department of Pharmacology University of Washington Seattle, Washington 98195

1. Historical Aspects of Protein Phosphorylation ......................... II. Protein Phosphorylation-Dephosphorylation Reactions A. General Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

B. Specificity for Phosphoryl Donors in Protein Kinase Reactions C. Protein Substrate Specificity of Protein Kinases and Phosphoprotein Phosphatases .................... ... D. The Reversibility of Protein K E. Autophosphorylation Reactions ................................. 111. Classification of Protein Kinases and Phosphoprotein Phosphatases . . . . . . . A. Protein Kinases ........................................ B . Phosphoprotein Phosphatases ................................... IV. Protein Phosphorylation and the Regulation of Biological Processes . . . . . . . A. Phosphorylation-Dephosphorylation versus Other Control Mechanisms ................... B. Diversity of Phosphorylatio Control Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References ........................................

1.

3 4 5

6 6 8 9 10

LO 14 15

15 17 18

Historical Aspects of Protein Phosphorylation

Of the many types of posttranslational modification of proteins that occur in cells relatively few are readily reversible. Those that are include acetylation, 3 THE ENZYMES, Vol. XVlI Copyright 0 1986 by Academic Press. Inc. All rights of repduction in any form reserved.

4

EDWIN G . KREBS

methylation, adenylylation, uridylylation, phosphorylation, and possibly one or two others such as ADP ribosylation. By all odds the most common type of reversible protein modification is phosphorylation, the process with which this volume is concerned. As has become abundantly clear nature has chosen phosphorylation-dephosphorylation as an almost universal mechanism for regulating the function of proteins, not only those that display enzymic activity but also proteins involved in many other biological processes. General recognition that protein phosphorylation has a major role in regulating protein function developed over a protracted period of time rather than being appreciated immediately as happened for allosteric regulation after the revelations by Monod et af. (1) in the mid-1960s. The first dynamic protein phosphorylation-dephosphorylation system to be elucidated was that involving glycogen phosphorylase, an enzyme that had been known to exist in two interconvertible forms, phosphorylases b and a (2, 3). In the mid-1950s these forms were shown to be nonphosphorylated and phosphorylated species of the eyzyme, which could be interconverted through the action of a protein kinase and a phosphoprotein phosphatase (4, 5 ) . In 1959 evidence was obtained that the kinase involved in this process, phosphorylase kinase, was itself regulated by phosphoxylation-dephosphorylation (6). A few years later it was determined that glycogen synthase also exists in interconvertible phosphorylated and nonphosphorylated forms (7). The fact that the first three phosphorylatable enzymes to be discovered all involved glycogen metabolism suggested to some that the process might be restricted to this area, but this idea was soon abandoned as further reports began to appear implicating enzymes that act in other pathways (8, 9). In particular, the studies by Lester Reed and his associates (10, 11)showing that pyruvate dehydrogenase is regulated by phosphorylation-dephosphorylation served to broaden people’s perspective regarding the scope of the process. At about this same time evidence was obtained that the enzyme that catalyzed the phosphorylation and activation of phosphorylase kinase was a cyclic AMP-dependent protein kinase that could catalyze the phosphorylation of other proteins, thus making it a probable mediator of the diverse actions of cyclic AMP (12, 13). These events served to stimulate general interest in the process of protein phosphorylation, and during the next fifteen years numerous enzymes were found to be regulated in this manner. In addition, many important functional proteins other than enzymes were shown to undergo reversible phosphorylation and to be controlled by this process. More detailed accounts of early work on protein phosphorylation have been reviewed elsewhere (14, 15).

II. Protein Phosphorylation-Dephosphorylation Reactions Phosphoproteins are formed in cells through the action of protein kinases, and they also occur as transient intermediates formed as part of the mechanism of

I.

5

ENZYMOLOGY OF CONTROL BY PHOSPHORYLATION

action of certain enzymes and transport proteins [reviewed in Ref. (16)]. It is with the former type of protein phosphorylation that the various chapters in this volume are concerned. A.

GENERALPROPERTIES

The reactions of protein phosphorylation-dephosphorylation involving protein kinases and phosphoprotein phosphatases are shown in Eqs. (1) and (2). Protein

+ nNTP-

Protein kinase

Protein-P,

+ nNDP

Phonphoprolein phosphalase

Protein-P,

+ nHzO

Protein

+ nPi

(1) (2)

As is implied in Eq. ( l ) , a given protein kinase often catalyzes the transfer of phosphate to more than a single site in its protein substrate. In this sense, glycogen phosphorylase, the first enzyme shown to undergo phosphorylationdephosphorylation and the prototype for many studies in this field, is unusual inasmuch as it is phosphorylated at a single site. In protein phosphorylation reactions it is common for a single protein to serve as the substrate for more than one kinase. In some instances different kinases catalyze the phosphorylation of identical sites in the protein substrate, but more commonly each kinase has its own site “preference” since, as discussed in Section II,C,2, protein kinases exhibit a moderately high degree of specificity for particular amino acid sequences surrounding phosphorylatable amino acid residues. The phosphorylation of an identical site by two or more protein kinases probably occurs due to a high degree of exposure of that site combined with the fact that protein kinases do not exhibit absolute specificities. The most intensively studied set of protein kinase reactions involving a single substrate are those that occur with glycogen synthase, which can be phosphorylated in vitro by no less than ten different protein kinases. It should be noted, however, that all of the kinases capable of phosphorylating glycogen synthase in virro may not function in this manner physiologically (see Chapters 11 and 12 in this volume). As with essentially all phosphotransferase reactions, protein kinase reactions require divalent metal ions, Mg2 probably serving as the physiologically significant cation in all instances, although manganous ions are nearly always effective in vifro. The actual substrate for the reaction depicted in Eq. (1) is the NTP-Me2+ complex. In some instances, however, a role for divalent ions in addition to forming the metal-nucleotide complex has been noted. This is seen, for example, with the cyclic AMP-dependent protein kinase, which binds free metal ions at a separate site (17, 18). The protein kinases constitute a very diverse set of enzymes, the total number of which is only now beginning to be appreciated. These enzymes are commonly regulated through their interaction with “second messengers” generated within cells in response to hormones and other extracellular agents (see below). The +

6

EDWIN G. KREBS

reactions catalyzed by the kinases are also regulated through the interaction of metabolites with their substrates, that is, through “substrate level” control (see, for example, Chapter 9, this volume and Chapter 2, Volume XVIII). The phosphoprotein phosphatases appear to be smaller in total number than the protein kinases, and they probably exhibit broader specificities. This in itself implies that these enzymes may not be as actively involved in regulating phosphorylation-dephosphorylation cycles as are the kinases, since the broader the specificity the greater the number of pathways that would be affected simultaneously by the regulation of a given enzyme. A more passive regulatory role for the phosphatases as compared to the kinases is also suggested by the fact that investigators have not discovered as many diverse forms of regulation for this set of enzymes. With respect to the last point, however, it is possible that the phosphatases have simply been more difficult to characterize than the kinases, and additional regulatory mechanisms may be found in the future. These matters are considered more fully in Chapter 8. B.

SPECIFICITY FOR PHOSPHORYL DONORS IN PROTEIN KINASEREACTIONS

The protein kinase reaction (Eq. 1) is written as being dependent on a nucleoside triphosphate, The physiologically significant donor in nearly all instances is probably ATP (19), although several protein kinases can also use GTP effectively in vitro. The last category includes casein kinase I1 [reviewed in Ref. (20)],a histone H1 kinase from tumor cells (21) and pp6CPrc (22). An unusual protein kinase from rabbit skeletal muscle, which utilizes phosphoenolpyruvate as a phosphate donor was reported by Khandelwal et al. (23). The kinase studied by this group was found to be activated by CTP (24). Phosphoenolpyruvate has long been known to serve as the phosphoryl donor in the phosphorylation of the enzyme involved in the first step of sugar transport in bacteria [reviewed in Ref. (25)],but this process belongs in a category set apart from typical protein kinase reactions as previously indicated. C.

PROTEINSUBSTRATE SPECIFICITY OF PROTEIN KINASES AND PHOSPHOPROTEIN PHOSPHATASES

1. Amino Acid Residues That Serve as Acceptors

of Phosphoryl Groups

Most of the acid-stable protein-bound phosphate found in cells is present as phosphoserine and phosphothreonine and is formed as a result of the action of protein serine and threonine kinases. A much smaller fraction, usually less than 0.2% of the total, is present as phosphotyrosine and arises as a result of the action of protein tyrosine kinases [reviewed in Ref. ( 2 6 ) ] .The possible existence of a

1 . ENZYMOLOGY OF CONTROL BY PHOSPHORYLATION

7

hydroxylysine kinase has also been indicated (27). In addition to protein kinases that catalyze the previously mentioned phosphorylations, there are indications that protein kinases exist that can catalyze the formation of acid-labile phosphohistidine and phospholysine in proteins (28-30). No work has been done on phosphatases that reverse the action of this last set of kinases, so it is not known whether such phosphorylations represent reversible protein modifications.

2 . Structural Determinants of Specificity The phosphorylation site sequences in various protein substrates for a number of different protein kinases have been determined and investigators using synthetic peptides as substrates have also elucidated a number of the structural requirements for substrate specificity of protein kinases. It has become clear that although the primary structure of a phosphorylation site sequence does not tell the whole story, it is usually possible to distinguish differences between the site specificities of the various kinases. For the CAMP-dependent protein kinase, for example, it can be predicted that if a given protein has an exposed -Arg-Arg-XSer-X- sequence, it will be phosphorylated by this enzyme. Similarly, casein kinase I1 phosphorylates exposed serine (or threonine) residues followed by one or more glutamic acids residues one position removed from the serine (e.g., -X-X-Ser-X-Glu-). The entire specificity pattern of a protein kinase is not revealed, however, simply by knowledge of its preferred primary amino acid sequence. Higher orders of protein structure also play a part in determining whether a given protein can be phosphorylated at an appreciable rate by a given kinase. These considerations are discussed in chapters that refer to specific kinases. The specificity of the phosphoprotein phosphatases is not as well understood as that of the kinases. One of the problems in studying this set of enzymes has been that earlier investigators were never certain whether their preparations contained more than one enzyme, and conclusions regarding specificity were obviously difficult to reach. Later studies, however, and particularly in the laboratory of Dr. Philip Cohen, made considerable headway in defining and classifying these enzymes. The relative activities of different phosphatases toward a number of different phosphoprotein substrates have been determined. Several general observations can be made: First, there is probably more overlapping of specificity among the phosphatases than is seen for the kinases. Second, it is evident that the pattern of specificity is not one in which a given phosphatase is “designed” to dephosphorylate a set of proteins phosphorylated by a given kinase. Thus, although the kinases recognize specific amino acid sequences in their substrates, this is not so readily apparent for the phosphatases. This suggests that higher orders of protein structure may be more important in governing substrate specificity for the phosphatases than for the kinases. If this is so, then one might anticipate that regulation of phosphatase activity might commonly

8

EDWIN G. KREBS

occur at the “substrate level” (i.e., through the interaction of metabolites with the phosphoproteins involved). Through such interactions and the ensuing conformational changes phosphatase activities could be regulated in a highly specific manner. Finally, it is apparent that some phosphatases exist that may attack lowmolecular-weight substrates as well as proteins under physiological conditions. Synthetic phosphopeptides have been employed only rarely in studies on phosphoprotein phosphatases. A pioneering effort in this regard was the work of Titanji et al. (31).

D. THEREVERSIBILITY OF PROTEIN KINASEREACTIONS The phosphorylation reactions that result in the formation of phosphoserine and phosphotyrosine in proteins, and presumably those that result in phosphothreonine, can all be demonstrated to be reversible albeit not under physiological conditions. Rabinowitz and Lipmann (32) were the first to describe the reversibility of protein phosphorylation. Using phosphorylated phosvitin as a phosphate donor, they demonstrated the formation of ATP from ADP in the presence of phosvitin kinase (probably casein kinase 11) from yeast or brain and postulated that runs of adjacent phosphoserines in phosvitin may have accounted for the apparent high free energy of hydrolysis of the bound phosphate in this protein. That such structural relationships are not essential for reversibility of protein kinase reactions was brought out by the work of Shizuta et al. (33), using the cyclic AMP-dependent protein kinase as enzyme and substrates in which no adjacent phosphoserines are present. These workers determined the equilibrium position in a well-defined protein kinase reaction and calculated that the free energy of hydrolysis of serine phosphate in this substrate (32P-labeled reduced carboxymethylated maleylated lysozyme) was -6.5 kcal mol- l . The reversibility of a number of other protein serine kinase reactions has been reported (34-37). Fukami and Lipmann (38) described reversibility of Rous sarcomaspecific immunoglobulin phosphorylation catalyzed by the src gene kinase and reported a free energy of hydrolysis of -9.48 kcal mol-I for protein-bound tyrosine phosphate, appreciably higher than the value of -6.5 kcal reported for protein serine phosphate by Shizuta et al. (33). However, the latter workers assumed a different AGO’ for hydrolysis of ATP (-8.4 kcal mol-l) than that used by Fukami and Lipmann (- 10 kcal mol- I ) . When the same value is used, a free energy of hydrolysis of -8.1 kcal mo1-I is obtained for protein serine phosphate (i.e., only slightly lower than that for tyrosine phosphate). A standard free energy of hydrolysis for serine phosphate in pyruvate kinase was found to be -6.6 kcal mol-I by El-Maghrabi et al. (37). These workers used -8.4 kcal mol-I for the free energy of hydrolysis of ATP. Very little has been made of the possible physiological significance of the reversal of protein kinase reactions, but it is possible that under some circumstances the process could be important. The fact that protein-bound serine phos-

1. ENZYMOLOGY OF CONTROL BY PHOSPHORYLATION

9

phate and tyrosine phosphate is relatively “energy rich” would indicate that a significant rate of turnover of this phosphate could occur in the absence of protein phosphatase activity. Potentially, the phosphate could be transferred to receptors other than proteins. In connection with the possible physiological significance of the reversal of protein kinase reactions, it is of interest that reversal of the phosphorylase kinase reaction required the presence of glucose or glycogen demonstrating that the reaction might be subject to regulation (36).

E. AUTOPHOSPHORYLATION REACTIONS Almost all protein kinases catalyze autophosphorylation reactions (i .e., reactions in which the kinase serves as its own substrate). Flockhart and Corbin (19) published a comprehensive list of the kinases that had been shown by the time of their review (1982) to exhibit this property. Their list could now be extended to include glycogen synthase kinase 3 (39), protein kinase C (40), the insulin receptor [reviewed in Ref. (41)] and a number of other oncogene-encoded protein tyrosine kinases (26). Autophosphorylation reactions can be intramolecular or intermolecular. This aspect was first examined with respect to the autophosphorylation of phosphorylase kinase, in which case the latter mechanism was shown to prevail (42). The autophosphorylation reaction involving type I1 cyclic AMP-dependent protein kinase, however, was found to occur by an intramolecular process (43).An intramolecular reaction was also found to be involved in autophosphorylationof the cyclic GMP-dependent protein kinase (44) and the insulin receptor (see Chapter 11). The significance of autophosphorylationreactions of protein kinases is readily apparent in some instances but not in all. With phosphorylase kinase autophosphorylation causes marked enhancement of activity, although there is no evidence that this is the physiologically significant process (see Chapter 10). Autophosphorylation of type II cyclic AMP-dependent protein kinase affects cyclic AMP binding and the dissociation-reassociation reactions of the enzyme in the presence of cyclic AMP [Ref. (45) and Chapter 31, an effect that may be of physiological significance, although again this has not been demonstrated. Of great interest is the fact that autophosphorylation of the insulin receptor renders this protein kinase independent of insulin (46). In some instances, particularly with those protein kinases in which autophosphorylation occurs by an intermolecular process and in which no functional change occurs in the activity of the kinase as a result of autophosphorylation, the reaction may simply be an expression of the lack of absolute specificity of protein kinases. In looking for autophosphorylation investigators often employ high concentrations of a kinase, and under these conditions autophosphorylation is observed even though the enzyme is actually an extremely poor substrate for itself. A possible unifying concept with respect to the significance of the auto-

10

EDWIN G. KREBS

phosphorylation reactions of protein kinase is that many of these enzymes may be kept in an inactive or inhibited state as a result of interaction between their autophosphorylation sites and their protein substrate binding sites. This phenomenon was first observed with type I1 cyclic AMP-dependent protein kinase. Here the autophosphorylation site on the regulation subunit (RI,), which possesses an amino acid sequence closely resembling a typical substrate sequence, is believed to interact with the active center of the catalytic subunit (see Chapter 3). Autophosphorylation of R,, renders it less potent as an inhibitor of R,, and facilitates activation by cyclic AMP. The latter activator, however, is capable of fully activating the enzyme with or without autophosphorylation. A similar, albeit not identical, situation appears to exist with the insulin receptor in which activation occurs as a result of an autophosphorylationreaction involving a site whose binding to the active center may well render the enzyme inactive in its dephospho form. The role of autophosphorylation in other protein tyrosine kinases is discussed in Chapters 6 and 7. For some protein kinases a functional autophosphorylation site may not exist (e.g., type I cyclic AMP-dependent protein kinase); instead, pseudoautophosphorylation sites are present. In these instances sites homologous to protein substrate phosphorylation sites are present but the sites lack phosphorylatable amino acid residues. For such enzymes activation is dependent on activators that alter conformation of the kinases and cause displacement of the pseudosubstrate sites from the active centers of these enzymes. Skeletal muscle myosin light chain kinase may be a second example of a protein kinase that is kept in its inactive form as a result of interaction of a pseudosubstrate site with the active center. This enzyme possesses what appears to be a site that shows strong homology to the phosphorylation site in myosin light chains. Interestingly, this site is close to or a part of the calmodulin binding site in this enzyme (47, 48).

111. Classification of Protein Kinases and Phosphoprotein Phosphatases

A. PROTEIN KINASES How many different protein kinases are there and how are they classified and named? The latter presents a difficult problem for a number of reasons. Classically, enzymes are grouped in accordance with the kind of chemical reaction that they catalyze and individual enzymes are then delineated and named on the basis of the specific substrate on which they act. This system breaks down for the protein kinases for several reasons. First, as previously indicated, a given protein kinase usually catalyzes the phosphorylation of a number of different proteins, so the selection of one particular substrate loses its meaning. Second, more often

I.


11

than not one protein can serve as a substrate for more than one kinase. At times, even the same phosphorylation site is involved with two or more kinases. Because of these complications, no systematic approach for naming this set of enzymes has been developed. In practice, it appears that protein kinases can be divided initially into broad classes based on whether the amino acid acceptor of the phosphoryl residue is (1) a protein alcohol group (i.e., the protein serine and protein threonine kinases), (2) a protein phenolic group (i.e., the protein tyrosine kinases), or (3) a protein nitrogen-containing group (i.e., the protein, histidine and lysine kinases). The existence of this last class of protein kinases is strongly suggested by the work of Roberts Smith and his collaborators (28, 29) and the report by Huebner and Mathews (30). Within each of the main classes of kinases individual groups of enzymes or subclasses exist, which more often than not can be delineated on the basis of the regulation of their activities. Table I presents a possible classification and nomenclature scheme for the protein serine and threonine kinases and Table I1 a scheme for the protein tyrosine kinases. Some of the groups or subclasses of protein serine and threonine kinases have only one known member at this time, but this situation may change. With respect to the grouping of the protein serine and threonine kinases shown in Table I, Group 6 consists of the so-called “independent” protein kinases (i.e., those that are not definitely known to be regulated through the interaction of a messenger(s) directly with the kinase). Some of the kinases in Group 6 are regulated as a result of the interaction of metabolites with their substrates. The individual protein serine and threonine kinases of Table I are not discussed here. The most intensively studied groups are the cyclic nucleotidedependent and calcium-calmodulin-dependent enzymes. Each of these groups is the subject of a separate chapter in this volume (see Chapters 3 and 4). The protein kinase that is regulated by diacylglycerol in the presence of calcium ions and phospholipid, protein kinase C, is also discussed in a separate chapter (see Chapter 5). One of the calcium-calmodulin (CaM)-dependent protein kinases, phosphorylase kinase, is itself an enzyme subject to regulation and is treated in the set of specific enzymes regulated by phosphorylation-dephosphorylation (see Chapter 10). Many of the protein serine and threonine kinases of Table I are discussed in chapters concerned with specific regulatory systems in which they function. For example, the double-stranded RNA-dependent and the hemininhibited protein kinases are handled in the chapter on the regulation of protein synthesis. Pyruvate dehydrogenase kinase and branched-chain ketoacid dehydrogenase kinase are discussed in Volume XVIII, Chapters 3 and 4 on the regulation of pyruvate dehydrogenase and branched chain keto acid dehydrogenase, respectively. The last set of enzymes (i.e., the two mitochondria1 dehydrogenases) are arbitrarily listed in Table I as independent kinases. They are both subject to regulation by metabolites that most logically would be thought to act by combin-

12

EDWW G. KREBS TABLE I

CLASSIFICATION AND NOMENCLATURE OF PROTEIN KINASESTHATCATALYZE THE TRANSFER OF PHOSPHATE TO PROTEIN ALCOHOL GROUPS(PROTEIN SERINE AND THREONINE KINASES) Group

Regulatory agent(s)

1

Cyclic nucleotides

2

Ca2+ and calmodulin

3

Diacylglycerol (Ca2+ and phospholipid) Double-stranded RNA Hemin Independent of regulation except at substrate level

4 5 6

Specific enzyme Type I cyclic AMP-dependent protein kinases Type I1 (H) cyclic AMP-dependent protein kinase“ Type I1 (B) cyclic AMP-dependent protein kinasea Cyclic GMP-dependent protein kinase Phosphorylase kinase (GSK 2)b Skeletal muscle myosin light chain kinase Smooth muscle myosin light chain kinase Multifunctional Ca2 + -calmodulin-dependent protein kinase Protein kinase C Double-stranded RNA-dependent protein kinase Hemin-inhibited protein kinase Fyruvate dehydrogenase kinase Branched-chain ketoacid dehydrogenase kinase Casein kinase I Casein kinase I1 GSK 3 (Factor FA)c GSK 4 Rhodopsin kinase HMG-CoA reductase kinase HMG-CoA reductase kinase kinase Histone HI kinase Histone kinase I1 H4-specific protease-activated protein kinase, Protease-activated kinase I1

“H” and “B” signify heart and brain isozymes. GSK stands for glycogen synthase kinase (see Chapter 11). Factor FA, name applied to a phosphoproteinphosphatase-activatingfactor that was later found to be a glycogen synthase kinase (49).

ing with their respective protein substrates. It should be noted, however, that acetyl-CoA, NADH, ADP, and pyruvate have all been reported to affect the activity of pyruvate dehydrogenase kinase when it is acting on a small synthetic peptide substrate, suggesting that the kinase itself interacts with these metabolites (see Volume XVIII,Chapter 3). Thus, it is possible that pyruvate dehydrogenase should be handled differently than shown in Table I (49). Some protein serine and threonine kinases have not been included in Table I because they have not been studied sufficiently to be certain of their status and relationship to other kinases. In this category the double-stranded DNA-depen-

1.


13

TABLE I1 CLASSIFICATION AND NOMENCLATURE OF PROTEIN KINASES THATCATALYZE THE TRANSFER OF PHOSPHATE TO PROTEIN PHENOLIC GROUPS(PROTEINTYROSINE KINASES) Group 1

2

Regulatory agent(s)

Specific enzymes

Polypeptide hormones EGF receptor and growth factors PDGF receptor Insulin receptor IGF-I receptor Unknown v-src and c-src-Encoded protein kinases v-yes and c-yes-Encoded protein kinases v-abl and c-abl-Encoded protein kinases v-fgr and c-fgr-Encoded protein kinases v-fps and c-fps-Encoded protein kinases v-fes and c-fes-Encoded protein kinases v-ros and c-ros-Encoded protein kinases

dent protein kinase (50) and the polypeptide-dependentprotein kinase (51) could be mentioned as examples. These and many new independent kinases will probably be implicated as having specific regulatory roles in the future. The protein tyrosine kinases have arbitrarily been divided into the two groups shown in Table I1 (i.e., the hormone or growth factor-dependent enzymes (see Chapter 7) and the independent kinases for which no regulatory agents are known (see Chapter 6). The latter set of kinases are encoded by certain retroviral oncogenes and their closely related cellular counterparts. Other independent protein tyrosine kinases not listed in Table II consist of the LSTRA kinase (52),a possibly related kinase from spleen (53),and TPK-75 from liver (54). The intriguing possibility that protein tyrosine kinases are regulated by intracellular second messengers is raised by reports suggesting that calcium-CaM and polyamines can stimulate tyrosine phosphorylation (55). The protein serine and tyrosine kinases all contain homologous catalytic domains indicating that they are members of a single family. This homology is shown in Fig. 1 in which the sequence of a portion (residues 49-253) of the catalytic subunit of beef cyclic AMP-dependent protein kinase (56) is used as a basis for calculating alignment scores (57) for comparing this sequence with the catalytic domains of other protein kinases. Comparisons are made with beef lung cyclic GMP-dependent protein kinase (58), the y-subunit of rabbit skeletal muscle phosphorylase kinase (59), rabbit skeletal muscle myosin light chain kinase (60),P 9 0 g a g - r u f (61), ppW" (62), the EGF receptor (63), and the insulin receptor (64).As can be seen there is a highly significant degree of homology for the catalytic domains in all of these protein kinases. As might be expected the

14

EDWIN G. KREBS

cAK cat subunit cGK

1

49

253

350

1

I

I

574

670

I

1

365 L

25

Phos K. Y subunit MLCK

pp6OSrc

I

302 I

20.0

386 I

17.0

I

273 I

12.0

510

603

I

A

480

526

I

718

Insulin Receptor Precursor

1017

/-

246

I

EGF Receptor

I

31.8

I

10.3

I

932 I

9.0

1242 I

1210 f P 1370 I

FIG. 1. Amino acid sequence homology between the catalytic domains of representative protein kinases. Each protein is represented by a horizontal bar, the length of which is proportional to the molecular weight of the protein as determined by its amino acid sequence (except for a portion of p90naS-raf). The small numbers indicate residue numbers. The large numbers indicate BarkerDayhoff alignment scores (57)of the catalytic domains to the various protein kinases compared to the segment between residues 49 and 253 in the catalytic subunit of the cyclic AMP-dependent protein kinase (cAK cat. subunit). cGK = cyclic GMP-dependent protein kinase; Phos. K = phosphorylase = chimeric protein containing raf oncogenekinase; MLCK = myosin light chain kinase; p9OnOg-..f encoded sequence and the gag sequence; pp6orrc = src-encoded protein tyrosine kinase.

degree of homology between the catalytic subunit of the cyclic AMP-dependent protein kinase and other protein serine and threonine kinases is higher than it is with the protein tyrosine kinase, except that this is not seen with respect to the oncogene-encoded protein p9ogng- ,f‘r which is thought to be a protein serine kinase (see Chapter 6). Phosphotransferases other than protein kinases have not been found to contain catalytic domains homologous to the catalytic subunit of the cyclic AMP-dependent protein kinase.

B. PHOSPHOPROTEIN PHOSPHATASES Early work on the phosphoprotein phosphatases was insufficient to make it feasible to classify or name them in a systematic manner. Impure preparations were often characterized and treated as single entities, when in fact more than one enzyme was present. What were thought of as different phosphatases often turned out to be dissociation products of more complex holoenzymes. In some instances “new” phosphatases were later recognized to be proteolytic artifacts.

1.


15

Fortunately, these enzymes that have been so recalcitrant to experimentation in the past are beginning to reveal their secrets. They are treated as a group in Chapter 8 and no attempt is made to classify them here.

IV. Protein Phosphorylation and the Regulation of Biological Processes A.

PHOSPHORYLATION-DEPHOSPHORYLATION VERSUS OTHERCONTROL MECHANISMS

Given the many different mechanisms that are available for regulating biological functions what is special about protein phosphorylation? Does this process have a unique role that cannot be carried out by other mechanisms? The last question is particularly pertinent when one considers the enormous potential for regulation through allosteric control involving the reversible binding of ligands to proteins. In fact, examination of the mechanisms for regulation of many of the specific enzymes described in this volume reveals that the kinetic parameters that are altered by ligand binding are usually the same as those changed by phosphorylation. Why attach a ligand (phosphate) covalently, when the reversible binding of metabolite can apparently accomplish the same purpose? (See Chapter 2, however, for a discussion of signal amplification by means of allosteric control versus cyclic covalent modification casades.) Reference to the role of protein phosphorylation in the regulation of glycogen phosphorylase may help to provide an answer to this question. In Chapter 9 Neil Madsen notes that conversion of phosphorylase b to phosphorylase a allows phosphorylase “to escape from the allosteric controls” to which it is subject when present in its b form; that is, after phosphorylation phosphorylase becomes fully active in the absence of 5‘-AMP and is no longer subject to inhibition by ATP or glucose 6-P. The implication is that there may be an advantage in having a system in which under certain circumstances an enzyme can be “frozen” in its active (or inactive) configuration regardless of what happens to metabolite levels. The same line of thinking may also apply to the purpose of a third form of regulation in which the amount of a given enzyme is altered through changes in its rate of biosynthesis or destruction. Dramatic changes in the concentration of an enzyme provides a means of superceding control by phosphorylation-dephosphorylation as well as allosteric control. Implicit in this form of reasoning is the often-voiced idea that allosteric control provides an immediate or almost instantaneous response to a given situation, control by protein phosphorylation provides an intermediate form of response on a time scale, and control through changes in the levels of particular proteins provides a slow but enduring response.

16

EDWN G . KREBS

IPROTEIN

PHOSPHORYLATION-DEPHOSPHORYLATION

1

FIG. 2. Protein phosphorylation and the transduction of hormonal signals.

Regardless of these considerations, although not entirely unrelated, is the fact that protein phosphorylation as a process is geared to the handling of signals that have an extracellular origin (hormones, growth factors, etc.), whereas allosteric control is utilized primarily for the handling of signals that arise intracellularly. The major pathways involving protein phosphorylation as a mechanism for relaying hormonal signals are shown in Fig. 2. One group of hormones, arbitrarily designated as Group 1, act through their receptors to raise or lower cyclic nucleotide levels in cells and thus affect the state of phosphorylation of cyclic nucleotide-dependent protein kinase substrates. A second large group of hormones and other extracellular agents (Group 2) trigger the breakdown of phosphatidylinositol diphosphate generating diacylglycerol and inositol triphosphate, the latter causing elevated intracellular calcium ion levels. The increase in calcium ions causes activation of a set of Ca2 -CaM-dependent protein kinases and one phosphoprotein phosphatase, leading to the modulation in the state of phosphorylation of protein substrates for these enzymes. The diacylglycerol formed as a result of the action of this group of hormones causes activation of protein kinase C, which again leads to protein phosphorylation. Finally, the third set of hormones (Group 3) act through receptors that possess protein kinase activity (i.e., protein tyrosine kinase activity). Although the nature of the protein substrates for this set of enzymes have not been fully elucidated (see Chapters 6 and 7) there appears to be little doubt that they constitute important phosphorylatable proteins involved in the action of these hormones. The same pro+

1 . ENZYMOLOGY OF CONTROL BY PHOSPHORYLATION

17

teins affected by the transmembrane signaling mechanism portrayed in Fig. 2, are also often targets for allosteric control through their interactions with metabolites generated within the cell. The concept that protein phosphorylation functions primarily to handle extracellular signals whereas allosteric control is utilized to handle intracellular signals is in keeping with the finding that eukaryotic cells contain much more phosphoprotein than is found in the prokaryotes (65, 66). This is not to say that protein phosphorylation is never used as a regulatory device in bacteria (see Volume XVIII, Chapter 14) but the extent to which it is employed is much less than in higher organisms in which appropriate response to extracellular signals is so important. On the other hand, allosteric control as a means of response to intracellular signals is highly developed in bacteria. It is of interest that whereas many protein kinases are regulated by second messengers generated in response to extracellular signals (Tables I and II and Fig. 2), relatively few of them are regulated by ordinary metabolites. The (at least partial) immunity of protein kinases to metabolite control may help to provide a mechanism whereby regulation by phosphorylation-dephosphorylation is segregated from allosteric control.

B. DIVERSITY OF PHOSPHORYLATION-DEPHOSPHORYLATION CONTROL MECHANISMS The existence of phosphoproteins first became known as a result of their nutritional significance. Thus, the early work in this field was concerned with the caseins of milk and the various phosphoproteins of egg yolk which serve as a source of phosphorus and amino acids for developing organisms. The second set of proteins to receive attention as targets for phosphorylation-dephosphorylation were the enzymes, many of which are reviewed in this volume. Concomitantly with work on enzyme phosphorylation-dephosphorylation, however, investigators extended their studies of this process to include a host of nonenzymic proteins involved in many different cellular functions, some of which (e.g., ion movements, receptor activity, and muscle contraction) are also treated here. Protein phosphorylation reactions are important in modulating nuclear events within the cell as is evidenced by the extensive work on histone and protamine phosphorylation as well as studies on the phosphorylation of nonhistone proteins. The exact nature of the role of protein phosphorylation reactions within the nucleus remains elusive, however, inasmuch as the precise function of the proteins involved is not always clear. An unusual role for phosphoproteins is that seen for a highly phosphorylated dentine protein that appears to function in the calcification process in teeth (67). Even under circumstances in which investigators are well aware of the physiological function of a particular protein known to be phosphorylated, they are not always able to find an effect of phosphorylation on the activity of these proteins.

18

EDWIN G. KREBS

Reactions in this category are often referred to euphemistically as “silent phosphorylations.” A possible explanation for such phosphorylations is that they may be targeting particular proteins for destruction. This possibility is discussed in Volume XVIII Chapters 2 and 7 in relation to pyruvate kinase and HMG-CoA reductase respectively; for each of these proteins phosphorylation causes the formation of a configuration that leads to enhanced susceptibility to proteases. A related role for protein phosphorylatin would be the part that it may play in the processing of certain proteins by proteases (68, 69). Other possible explanations for silent phosphorylations should also be considered. Protein-protein interactions within the cell may be regulated by phosphorylation-dephosphorylation. The intracellular localization of proteins may be influenced by phosphorylation. These and many other protein functions are difficult to measure outside of the natural intracellular environment. Finally, with respect to silent phosphorylations, it is possible that some protein phosphorylation-dephosphorylationsserve no particular purpose other than generating heat and simply constitute futile cycles. REFERENCES I. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Monod, J . , Changeux, J.-P., and Jacob, F. (1963). JMB 6, 306-329. Con, G. T., and Green, A. A. (1943). JBC 151, 31-38. Con, G. T., and Cori, C. F. (1945). JBC 158, 321-332. Fischer, E. H., and Krebs, E. G. (1955). JBC 216, 121-132. Sutherland, E. W., and Wosilait, W. D. (1955). Nurure (London) 175, 169-171. Krebs, E. G., Graves, D. J., and Fischer, E. H. (1959). JEC 234, 2867-2873. Friedman, D. L., and Lamer, J . (1963). Biochemistry 2,669-675. Rizak, M. A. (1964). JBC239, 392-395. Mendicino, J . , Beaudreau, C., and Bhattachryya, R. N. (1966). AEB 116,436-445. Linn, T. C., Pettit, F. H., Hucho, F., and Reed, L. J. (1969). PNAS 64, 227-234. Linn, T. C., Pettit, F. H., and Reed, L. J. (1969). PNAS 62, 234-241. Walsh, D. A . , Perkins, J. P., and Krebs, E. G. (1968). JBC 243, 3763-3765. Kuo, J. F., and Greengard, P. (1969). JBC 244, 3417-3419. Krebs, E. G. (1983). Philos. Trans. R. Soc. London, Ser. B 302, 3-11. Krebs, E. G. (1985). Trans. Biochem. Soc. 13, 813-820. Krebs, E. G. (1972). Curr. Top. Cell. Regul. 5 , 99-133. Armstrong, R. N., Kando, H., Geanot, J., Kaiser, E. T., and Mildvan, A. S. (1979). Biochemisrry 18, 1230-1238. Hixson, C. S . , and Krebs, E. G. (1979). JBC 254, 7509-7514. Flockhart, D. A., and Corbin, J. D. (1982). CRC Crir. Rev. Eiochem. 13, 133-186. Hathaway, G. M . , and Traugh, J. A. (1982). Curr. Top. Cell. Regul. 21, 101-127. Quirin-Steicker, C., and Schmitt, M. (1981). EJB 118, 165-172. Graziani, Y., Erikson, E., and Enkson, R. L. (1983). JBC 258, 6344-6351. Khandelwal, R. L., Mattoo, R. L., and Waygood, E. B. (1983). FEBSLerr. 162, 127-132. Mattoo, R. L., Waygood, E. B., and Khandelwal, R. L. (1984). FEBS Len. 165, 117-120. Simoni, R. D., and Postma, P. W. (1975). Annu. Rev. Biochem. 44, 523-554. Hunter, T., and Cooper, J. A. (1985). Annu. Rev. Eiochem. 54, 897-930.

1.


19

27. Urishizaki, Y., and Seifter, S. (1985). Biophys. J . 47, 233a. 28. Smith, D. L., Chen, C. C., Bruegger, B. B., Holtz, S. L., Halpern, R. M., andsmith, R. A. ( I 974). Biochemistry 13, 3780-3785. 29. Smith, R. A,, Halpern, R. M., Bruegger, B. B., Dunlop, A. K., and Fricke, 0. (1978). Merhods Cell Biol. 14, 153-159. 30. Huebner, V. D., and Mathews, H. R. (1985). FP 44, Abstr. 3882. 31. Titanji, V. P. K., Ragnarsson, U., Humble, E., and Zetterquist, 0. (1980). JBC 255, 1133911343. 32. Rabinowitz, M., and Lipmann, F. (1960). JBC 235, 1043-1056. 33. Shizuta, Y., Beavo, J. A., Bechtel, P. J., Hofmann, F., and Krebs, E. G. (1975). JBC 250, 6891 -6896. 34. Lerch, K., Muir, L. W., and Fischer, E. H. (1975). Biochemisrry 14, 2015-2023. 35. Rosen, 0. M., and Erlichman, J. (1975). JBC 250, 7788-8894. 36. Shizuta, Y . , Khandelwal, R. L., Maller, J. L., Vandenheede, J. R., and Krebs, E. G. (1977). JBC 252, 3408-3413. 37. El-Maghrabi, M. R., Haston, W. S., Flockhart, D. A , , Claus, T. H., and Pilkis, S. J. (1980). JBC 255, 668-675. 38. Fukami, Y., and Lipmann, F. (1983). PNAS 80, 1872-1876. 39. Hemmings, B . A , , Yellowlees, D., Kernohan, J. C., and Cohen, P. (1981). EJB 119,443-451. 40. Kikkawa, U., Takai, Y., Minakuchi, R., Inohara, S., and Nishizuka, Y. (1982). JBC 257, 13341- 13348. 41. Kahn, C. R., White, M. F., Grigorescu, F., Takayarna, S., Haring, H. U., and Crettaz, M. (1985). In “Molecular Basis of Insulin Action” (M. P. Czech, ed.), pp. 67-93. Plenum, New York. 42. DeLange, R. J . , Kemp, R. G., Riley, W. D., Cooper, R. A,, and Krebs, E. G. (1968). JBC 243, 2200-2208. 43. Rangel-Aldao, R., and Rosen, 0. M. (1976). JBC 251, 7526-7529. 44. Lincoln, T. M., Flockhart, D. A , , and Corbin, C. D. (1978). JBC 253, 6002-6009. 45. Erlichman, J . , Rosenfeld, R., and Rosen, 0. M. (1974). JBC 249, 5000-SO03. 46. Rosen, 0. M., Herrara, R., Olowe, Y., Petruzzeli, L. M., and Cobb, M. (1983). PNAS 80, 3237-3240. 47. Blumenthal, D. K., Takio, K., Edelman, A. M., Charbonneau, H., Titani, K., Walsh, K. A., and Krebs, E. G. (1985). PNAS 82, 3187-3191. 48. Edelman, A. M., Takio, K., Blumenthal, D. K., Hansen, R. S., Walsh, K. A., Titani, K., and Krebs, E. G. (1985). JBC 260, 11275-11285. 49. Vandenheede, J. R., Yang, S. D., Goris, J., and Merlevede, W. (1980). JBC 255, 1176811774. 50. Walker, A. I . , Hunt, T., Jackson, R. H., and Anderson, C. W. (1985). EMBO J . 4, 139-145. 51. Racker, E., Abdel-Ghany, M., Sherril, K., Riegler, C., and Blair, E. A. (1984). PNAS 81, 4250-4254. 52. Casnellie, J . E., Gentry, L. E., Rhorschneider, L. R., and Krebs, E. G. (1984). PNAS 81, 6676-6680. 53. Swamp, G., Dasgupta, J . D., and Garbers, D. L. (1983). JBC 258, 10341-10347. 54. Goldberg, A. R., and Wong, T. W. (1984). Adv. Enzyme Regul. 22, 289-308. 5 5 . Migliaccio, A,, Rotondi, A,, and Auricchio, F. (1984). PNAS 81, 5921-5925. 56. Shozo, S., Parmelee, D. C., Wade, R.D., Kumar, S., Ericsson, L. H., Walsh, K. A., Neurath, H., Long, G. L., Demaille, J. G., Fischer, E. H., and Titani, K. (1981). PNAS 78, 848-851. 57. Barker, W. C., and Dayhoff, M. 0. (1982). PNAS 79, 2836-2839. 58. Takio, K., Wade, R. D., Smith, S. B., Krebs, E. G., Walsh, K. A,, and Titani, K. (1984). Biochemisrry 23, 4207-4218.

20

EDWIN G. KREBS

59. Reimann, E. M., Titani, K., Ericsson, L. H., Wade, R. D., Fischer, E. H., and Walsh, K. A. (1984). Biochemistry 23, 4185-4192. 60. Takio, K., Blumenthal, D. K., Edelman, A. M., Walsh, K. A,, Krebs, E. G., and Titani, K. (1985). Biochemistry 24 (in press). 61. Mark, G. E., and Rapp, U. R. (1984). Science 224, 285-289. 62. Czernilofsky, A. P., Levinson, A. D., Varmus, H. E., Bishop, J. M., Tischer, E., and Goodman, H. (1980). Nature (London) 287, 198-203. 63. Ullrich, A., Cousens, L., Hayflick, J. S., Dull, T. J., Gray, A., Tam, A. W.,Lee, J., Yarden, Y.,Libermann, T. A., Schlessinger, J., Downward, J., Mayes, E. L. V., Whittle, N., Waterfield, M. D., and Seeburg, P. H. (1984). Nature (London) Jo9, 418-425. 64. Ullrich, A., Bell, J. R., Chen, E. Y.,Herrera, R., Petruzzelli, L. M., Dull, T. J., Gray, A,, Coussens, L., Liao, Y .-C., Tsubokawa, M., Mason, A., Seeburg, P. H., Grunfeld, C., Rosen, 0. M., and Ramachandran, J. (1985). Nature (London) 313, 756-761. 65. Sedmak, J., and Ramaley, R. (1968). BBA 170, 440-442. 66. Forsberg, H., Zetterqvist, O., and Engstrom, L. (1969). BBA 181, 171-175. 67. Pal, B. K., and Roy-Berman, P. (1978). BBRC 81, 344-350. 68. Kaufman, J. F., and Strominger, J. L. (1979). JBC 76, 6304-6308. 69. Veis, A., and Perry, A. (1967). Biochemistry 6, 2409-2416.

Cyclic Cascades and Metabolic Regulation EMILY SHACTER* P. BOON CHOCK SUE GOO RHEE EARL R. STADTMAN Laboratory of Biochemistry National Heart, Lung, and Blood Institute National Institutes of Health Bethesda, Maryland 20892

1. Perspectives ................................... ...... 11. Features of Cyclic Cascade Systems . . . . . . . . . . . . . . . A. Unidirectional versus Interconvertible Cascades .................... B. Theoretical Analysis .......................................... 111. Experimental Verification of the Cyclic Cascade Model ................ IV. Energy Consumption ...... V. Covalent Interconversion ve VI. Concluding Remarks ............................................. ........ References .............

1.

21 27 28 33 37 37

Perspectives

Reversible covalent modification of an interconvertible protein involves the enzymic transfer and removal of a modifying group from a donor molecule to a specific amino acid residue. The different forms of reversible covalent modification are listed in Table I (1-14). Each of these modifications and demodifications *Resent address: Laboratory of Genetics, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 21 THE ENZYMES,Vol. XVII Copyright 8 1986 by Academic Press, Inc. All rights of repduction in any form reserved.

22

SHACTER, CHOCK, RHEE, AND STADTMAN TABLE I REVERSIBLE COVALENT MODIFICATIONS OF PROTEINS Modification

Donor molecule

Phosphorylation

ATP, GTP

ADP-ribosylation

NAD +

Nucleotidylylation (adenylylation and uridylylation) Methylation

ATP, UTP

Acetylation Tyrosylation Sulfation

S-Adenosyl-methionine

Acetyl-CoA Tyrosine 3-Phosphoadenosine 5-phosphosulfate

Amino residue(s)

References

Serine Threonine Tyrosine Hydroxylysine Arginine Glutamate Lysine (terminal COOH) Diphthamide Tyrosine Serine Aspartate Glutamate Lysine Histidine Glutamine Lysine Carboxyl terminus Tyrosine

is catalyzed by specific converter enzymes such as protein kinases and phosphoprotein phosphatases. The unification of two opposing converter enzymes and an interconvertible enzyme forms a cyclic cascade system. Thus, any enzyme or protein that undergoes reversible covalent modification is a member of a cyclic cascade. The number of enzymes and pathways regulated by cyclic cascade systems has increased dramatically since the discovery in 1955 by Sutherland and Wosilait (15) and by Fischer and Krebs (16) that glycogen phosphorylase exists in two forms: an active phosphorylated form and a relatively inactive unmodified form. A comprehensive list of interconvertible enzymes and proteins is given in Table I1 (17-159).Proteins that undergo covalent modification but that have not been are not included. functionally defined, such as miscellaneous lens proteins (160), Nonetheless, Table I1 is almost five times as long as a similar list published in 1978 (161) and its volume continues to increase. It should be pointed out that tyrosine sulfation is not known to be cyclic nor has it been shown to regulate any enzyme activity. However, in light of evidence that the membranal receptor for interleukin-2 contains tyrosine sulfate (88) and other results showing that the level of tyrosine sulfate decreases in cells that are transformed by retroviruses

2. CYCLIC CASCADES AND METABOLIC REGULATION

23

TABLE I1 INTERC0NVtRTIBI.E

ENZYMG (PRO T t l N S )

Enzyme

Rcferences

Phosphorylution Acetyl-CoA carboxylase Acetylcholine receptor (nicotinic) Actin Acyl coenzyme A: cholesterol acyltransferase Alpha-ketoacid dehydrogenase complex Aminoacyl-tRNA synthetase Angiotensin Asialoglycoprotein receptor ASpdrdginaSe ( L . WiichOtii) ATP citrate lyase Beta-adrenergic receptor Beta-glucuronidase Branched-chain 2-oxoacid dehydrogenase complex Ca2 -ATPase C-protein (cardiac muscle) Casein CAMP-dependent protein kinase (regulatory subunit) CAMP-dependent protein kinase (catalytic subunit) cGMP-dependent protein kinase Cholesterol esterase Cholesterol 7a-hydroxylase Cyclic nucleotide phosphodiesterase Cytochrome P-450 (adrenocortical) DARPP-32 DNA topoisomerase I DNA topoisomerase 11 Enolase Epidermal growth factor receptor Erythrocyte membrane band 3 protein Eukaryotic intiation factor-2 (eIF-2) Eukaryotic initiation factor-3 (eIF-3) Eukaryotic peptide elongation factor- I Fibrinogen Fodrin (nonerythroid spectrin) Fructose- I ,6-bisphosphatase Fructose-:! ,6-bisphosphatase-fructose-6-P-2-kinase y-Aminobutyric acid (GABA) receptor complex Glucocorticoid receptor (liver) Glutamate dehydrogenase, NAD-dependent (yeast) Glycerol phospate acyltransferase Glycogen phosphorylase Glycogen synthase G-substrate (brain) +

(continued)

24

SHACTER, CHOCK, RHEE, AND STADTMAN TABLE 11 (Conrinued) Enzyme Guanylate cyclase High-mobility group (HMG) proteins Histones HLA antigens (Class I) Hormone-sensitive lipase-diglyceride lipase 3-Hydroxy-3-methylglutarylcoenzyme A reductase Hydroxymethylglutaryl coenzyme A reductase kinase Immunoglobulin G Insulin receptor Interleukin-2 (IL-2, T-cell growth factor) receptor lsocitrate dehydrogenase Keratin (Type I) Lactate dehydrogenase Microtubule-associated protein-2 (MAP-2) Myelin basic protein Myeloperoxidase Myosin heavy chains Myosin light chains Myosin light chain kinase Na+ channel, a subunit Na+ -K -ATPase Omithine decarboxylase 3-0x0-Sa-steroid A4-dehydrogenase Phenylalanine hydroxylase Phosphofructokinase Phosphoglycerate mutase Phospholamban Phospholipid methyltransferase Phosphoprotein phosphatase inhibitor- 1 Phosphoprotein phosphatase inhibitor-2 Phosphorylase kinase Phosvitin Platelet-derived growth factor receptor Poly(A) polymerase Polyoma T antigen Prolactin Protamine Pyruvate dehydrogenase Pyruvate kinase Pyruvate, Pi, dikinase (lean Retroviral oncogenes Reverse transcriptase Rhodopsin Ribosomal protein S6 RNA Polymerase (DNA-dependent) Synapsin I (brain) +

References

25

2. CYCLIC CASCADES AND METABOLIC REGULATION TABLE I1 (Continued) Enzyme Tau factor Troponins I and T Tubulin Tyrosine hydroxylase Vinculin ADP-ribosylation Actin Adenylate cyclase Ca2+ ,Mg2+ -dependent endonuclease DNA ligase I1 DNA polymerases a and p DNA topoisomerase I Elongation factor I1 (E. coli) Glutamine synthetase (mamm.) High-mobility group proteins Histones Micrococcal nuclease Nitrogenase (bacterial) Phosphorylase kinase (skeletal muscle) Poly(ADP-ribose) synthetase Protamines RNA polymerase (E. coli) RNase (bovine) SV40 T antigen (large) Terminal deoxynucleotidyl transferase Transducin Nucleofidylylafion Aspartokinase (E. coli) Glutamine synthetase ( E . coli) Regulatory protein, PI, ( E . coli) SV40 T antigen (large) Methylation ACTH Actin Calmodulin Citrate synthase Cytochrome c EF-TU y-Globulin High-mobility group (HMG) proteins -1 and -2 Histones Methyl-accepting chemotaxis proteins (MCPs) Myosin Opsin Ovalbumin Ribonuclease

References

(153) (8) (8) (9) (154)

(10) (154)

(10) (11)

(10) (154) (10) (21)

(155) (1555)

(10) (154) ( 154 1

26

SHACTER, CHOCK, RHEE, AND STADTMAN TABLE I1 (Continued) Enzyme

References

Acetylation High-mobility group (HMG) proteins Histones Tyrosinolation Tubulin a-subunit Sulfation Complement C4 Fibrinogens, fibrins Gastrin I1 Phyllokinin Caerulein Cholecystokinin Leu-enkephalin Fibronectin Immunoglobulins Interleukin-2 (IL-2, T-cell growth factor) reCeptor

(158), it seems possible that sulfation may have a regulatory function and, therefore, it has been included in the tables. Perusal of Table I1 reveals that a broad array of cellular pathways, such as protein, carbohydrate, lipid, and nucleic acid metabolism; interferon action (162); DNA repair (163); viral oncogenesis (164-166); muscle contraction (98); and membranal signal transduction are regulated by cyclic covalent modification. In addition, a number of regulatory enzymes undergo multiple covalent modifications. For example, glycogen synthase is phosphorylated on seven different sites by five different protein kinases (108),each of which is regulated by different stimuli and effectors. Furthermore, some proteins, such as phosphorylase kinase, histones, and the interleukin-2 receptor undergo more than one form of modification. Such multisite modification not only expands the potential of regulatory networks numerically, but also creates the possibility for competition between different converter enzymes and, hence, between different cyclic cascade systems. This has been suggested specifically in the case of ADPribosylation and phosphorylation of phosphorylase kinase and miscellaneous nuclear proteins (148, 167). Cyclic cascade systems respond to metabolic requirements by shifting an interconvertible protein between different extents of modificakon. In the case of an interconvertible enzyme, this produces a change in its specific activity either by increasing or decreasing the K , or V,,, of the enzyme or by modulating its response to allosteric effectors. For example, phosphorylation of glycogen phosphorylase results in activation of the enzyme by increasing the V,,,, whereas phosphorylation of myosin light chain kinase by CAMP-dependent protein kinase


27

inhibits the enzyme by decreasing its affinity for the essential activation complex, Ca2+-calmodulin. Analogously, in the case of a binding protein (e.g., eIF-2, histones), covalent modification can either enhance or diminish binding efficiency. Key to their pivotal role in cellular regulation is the fact that cyclic cascade systems are regulated by allosteric interactions of metabolic effectors with either the converter enzymes or the interconvertible enzymes. Moreover, as previously mentioned, covalent modification can modulate the allosteric interactions between metabolic effectors and an interconvertible enzyme. Thus, these two major mechanisms of enzyme regulation are inextricably intertwined [see Ref. (168) for a review of the multimodulation of enzyme activity]. Covalent modification has also been implicated in regulating specific protein degradation in the cell. That is, phosphorylation of several regulatory enzymes has been shown to increase their susceptibility to proteolysis both in virro and in vivo; these include HMG-CoA reductase (169), pyruvate kinase (170), glutamate dehydrogenase from yeast, (171), fructose-1,6-bisphosphatase (172), and cardiac troponin (173). Similarly, ubiquitination of proteins (which is not known to be cyclic) has been proposed to be involved in marking proteins for ATPdependent proteolysis (174). Thus, covalent modification may be a mediator for the third major mechanism of metabolic regulation-that of controlling intracellular enzyme levels. Clearly, a thorough understanding of cyclic cascades is requisite to our ability to understand and manipulate normal and abnormal cell function. The following sections present a brief update and summary of the characteristics of cyclic cascade systems that make them so well suited for their central position in cellular regulation. More detailed reviews can be found in Refs. (8) and (161).

II. Features of Cyclic Cascade Systems A.

UNIDIRECTIONAL VERSUS INTERCONVERTIBLE CASCADES

Covalent modifications of enzymes (proteins) is catalyzed by converter enzymes such as kinases and phosphatases. They involve the action of one enzyme upon another and are therefore referred to as cascade systems (8, 161, 175). These systems can be divided into two classes-unidirectional cascades and cyclic cascades. Unidirectional cascades are irreversible and are usually involved in proteolytic cleavage of a specific peptide bond, as occurs in the activation of zymogens (176). Well-recognized unidirectional cascades include the bloodclotting cascades (177) and the cascade involved in complement fixation (178). They are designed as amplifiers that, in response to certain alarm signals, generate an avalanche of product required to meet specific biological challenges. When the need suhsides the cascade are terminated. Therefore, unidirectional

28

SHACTER, CHOCK, RHEE, AND STADTMAN

cascades are contingency systems serving as biological switches that can be turned ON to meet occasional emergency situations. In contrast, cyclic cascades involve the derivatization of one or more specific amino acid residues within the protein as occurs in the ATP-dependent phosphorylation of the hydroxyl groups of serines, threonines, or tyrosines ( I ) , in the ATP-dependent adenylylation of the hydroxyl group of a specific tyrosine in E. coli glutamine synthetase (179, 180), and others described in Table I. These enzyme derivatizations are cyclic processes resulting from the coupling of two opposing cascades-one concerned with the covalent modification and the other with the demodification of an interconvertible enzyme. The properties of cyclic cascades elucidated by the theoretical analysis discussed in the following section show that they are endowed with unique characteristics that provide the cell with the capacity to regulate and coordinate a multitude of metabolic pathways. B. THEORETICAL ANALYSIS In view of the vast number of interconvertibleproteins listed in Table 11, many of which are key enzymes in metabolic control, we have carried out a theoretical analysis designed to reveal the advantages of such a complex mechanism for cellular regulation (161, 181). In this analysis, it has been ascertained that the covalent modification of enzymes does not function simply as an ON-OFF switch for various metabolic pathways, but rather that it is part of a dynamic process in which the fractional activities of the interconvertible enzymes can be varied progressively over a wide range. This concept derived from the experiments of Brown et al. (182) demonstrating that the adenylylation of glutamine synthetase is not an all-or-none process; instead, a steady state is established and its level is modulated by the concentrations of effectors involved. Similar observations have been reported by Pettit et al. (183) for the mammalian pyruvatedehydrogenase complex. The cyclic cascade model, as depicted in Fig. 1 , derives from the coupling of a forward cascade and a reverse cascade. The forward cascade involves the activation of the inactive converter enzyme, PK, by an allosteric effector, e,. The activated converter enzyme, PK,, then catalyzes the modification (i.e., phosphorylation) of the interconvertible enzyme from its unmodified form, S, to its modified form, S-P. In the reverse cascade, the inactive converter enzyme, PT, is activated by an allosteric effector, e2, and the active converter enzyme catalyzes the demodification (i.e., dephosphorylation)of the modified interconvertible enzyme, S-P, to its unmodified form. The coupling of these two cascades results from the fact that the substrate of the converter enzyme in one cascade is the product of the opposing cascade. Note also that for each complete cycle, a molecule of ATP is hydrolyzed to ADP and Pi. When the ATP concentration is maintained in excess relative to the enzymes involved and at a fairly constant

29

2. CYCLIC CASCADES AND METABOLIC REGULATION PK

+

el

K1 + PK,

I I I

ADP

ATP

PT

+ 82 +PT, K2

FIG. 1 . Schematic representation of a monocyclic cascade. K,,KZ,K l f , KI,are dissociation constants for PK,, P",, S-PK,, and S-P.FTa,respectively; klf and k I r are specific rate constants for the reactions designated.

level, a steady state is established in which the rate of S-P formation is equal to the rate of S regeneration. Using the parameters shown in Fig. 1, it is possible to derive a relatively simple equation to quantitate the fractional phosphorylation of the interconvertible enzyme (161, 181). When the modified interconvertible enzyme in one cycle catalyzes the modification of an interconvertible enzyme in another cycle, the two cycles are coupled such that the fractional modification of the second interconvertible enzyme is a function of all the quantitative parameters that define both cycles. Similarly, when the cascade is composed of n interconvertible enzymes and the modified interconvertible enzyme in one cycle functions as the converter enzyme for the next interconvertible enzyme, a multicyclic cascade composed of n cycles is obtained (161, 184). Quantitative analysis of cyclic cascades reveals (8, 161, 182, 184) that ( a ) they are endowed with an enormous capacity for signal amplification. As a result, they can respond to primary effector (e, in Fig. 1) concentrations well below the dissociation constant of the effector-enzyme complex; ( b ) they can modulate the amplitude of the maximal response that an interconvertible enzyme can accomplish at saturating concentrations of allosteric effectors; ( c ) they can enhance the sensitivity of modification of the interconvertible enzyme to changes in the concentrations of allosteric effectors (i.e., they are capable of eliciting apparent positive and negative cooperativity in response to increasing concentrations of allosteric effectors); (d)they serve as biological integration sysrems that can sense simultaneous fluctuations in the intracellular concentrations of numerous metabolites and adjust the specific activity of the interconvertible enzymes

30


accordingly; (e) they are highly flexible with respect to allosteric regulation and are capable of exhibiting a variety of responses to primary allosteric stimuli; and v) they serve as rule amplifiers and therefore are capable of responding extremely rapidly to changes in metabolite levels (185, 186). 1. Signal Amplification This is a time-independent parameter defined (161, 181) as the ratio of the concentration of the primary allosteric effector (el in Fig. 1) required to attain 50% activation of the converter enzyme PK to the concentration required to produce 50% modification of the interconvertible enzyme S. This exceptional property derives from the fact that enzymes (catalysts) act as intermediaries between the metabolic effectors (signals) and the target interconvertible enzymes. Signal amplification can be quantified using the steady-state expression for the fractional modification of the interconvertible enzyme. It is a multiplicative function of the variables shown in Fig. 1; each of these parameters is susceptible to modulation by allosteric effectors. Because of the multiplicative nature of the fractional modification expression, small changes in several of the parameters can lead to enormous changes in fractional modification of the intercovertible enzyme in response to effector. Moreover, signal amplification increases exponentially as a function of the number of cycles in the cascade. Due to signal amplification, interconvertibleenzymes can respond to effector concentrations that are well below the dissociation constant of the effector-converter enzyme complex. In other words, only a relatively small fraction of converter enzyme need be activated in order to obtain a significant fractional modification of the interconvertible enzyme. It should be pointed out that the signal amplification described here is distinctly different from catalytic amplification, which is solely a function of the expansion of the relative concentrations and catalytic efficiencies of the converter and interconvertible enzymes. For the unique case in which the maximal specific catalytic activities of the converter enzyme, PK, and the interconvertible enzyme, S, are the same, the catalytic amplification potential is equal to the concentration ratio [S]/[PK]. In fact, in most cascades that have been studied, there exists a pyramidal increase in the concentration of the cascade enzymes; that is, the concentration of converter enzymes is significantly lower than that of the interconvertible enzyme substrate. Therefore, they possess a high catalytic amplification potential. This enhances the signal amplification of the cascade. 2 . Amplitude Amplitude is defined (161) as the maximal value of fractional modification of the interconvertible enzyme attainable with saturating concentrations of an effector. By changing the magnatude of the cascade parameters, the amplitude can change smoothly from nearly 100% to nearly 0%. Therefore, even at saturating


31

levels of an effector (e.g., e , in Fig. 1) interconvertibleenzymes do not function as ON-OFF switches. Gresser has pointed out (1860) that all modification reactions would proceed to their individual equilibrium. For physiological cascades, e.g., phosphorylaton/dephosphorylation,adenylylation/deadenylylation, etc., almost all interconvertible proteins would be present in only one form when one considers the individual equilibria for either the modification or demodification reaction. Thus, under certain extreme conditions, such as the temporary loss of either the regeneration cascade or forward cascade, enzyme cascades can effectively function as ON-OFF switches for enzymic activity.

3 . Sensitivity Cyclic cascades can generate either apparent positive or negative cooperative responses of fractional modification (i.e., enzymic activity) of the interconvertible enzyme to increasing concentrations of an allosteric effector (161). These apparent cooperativities can derive from the synergistic or antagonistic effects that a single allosteric effector exerts on two or more steps in the cascade. Thus, a sigmoidal response need not reflect positive cooperativity in the binding of an effector to multiple binding sites on the converter enzyme. Instead, it can be accomplished when an effector activates the forward converter enzyme and inactivates the reverse converter enzyme, or vice versa. Consequently, an effective way for obtaining high sensitivity is to have both forward and reverse converter enzyme activities combined in a single polypeptide such that binding of one effector can activate one activity while inactivating the other activity. Four such bifunctional enzymes have been isolated and characterized. They are the uridylyltransferase-uridylyl-removing enzyme activities that catalyze the uridylylation-deuridylylation of the P,, regulatory protein in the E. coli glutamine synthetase cascade (183, the adenylyltransferase that catalyzes the adenylylation-deadenylylation of glutamine synthetase (188, 189), a protein kinasephosphatase that catalyzes the phosphorylation and dephosphorylation of isocitrate dehydrogenase in E. coli (190), and the 6-phosphofructose-2-kinase-fructose-2, 6-bisphosphatase that catalyzes the synthesis and breakdown of fructose 2, 6-bisphosphate (191, 192). All four of these bifunctional enzymes are involved in cyclic cascade systems. A sigmoidal response can also be achieved by the regulatory mechanism reported by Jurgensen (193) in which the type I1 regulatory subunit of CAMPdependent protein kinase inhibits the MgATP-dependent protein phosphatase. In addition, apparent cooperativity can be obtained when the active converter enzyme forms a tight complex with the interconvertible enzyme as shown by Shacter e f al. (194, 195) and independently by Goldbetter and Koshland (196). The latter authors showed that when the converter enzyme is saturated by its interconvertible substrate, the cyclic cascade can exhibit a sigmoidal response to effector concentration. They call this effect “zero-order ultrasensitivity.”

32


4. Flexibility and Biological Integration There are two aspects of flexibility that need to be considered in discussing the properties of cyclic cascades; namely, the flexibility for generating various allosteric control patterns and the flexibility in regulation by multiple metabolites. The control pattern depicted in Fig. 1 illustrates just one of many variations that can be derived by changing the nature of the interactions between the allosteric effectors, e l and e2, and the converter enzymes, PK and PT. For the case in which the converter enzyme PK is activated by effector e l , four different regulatory mechanisms may result depending upon whether e, and e2 activate or inhibit the activity of converter enzyme PT (181). Numerical analysis of these four mechanistic schemes shows that they can yield an array of patterns of fractional modification of the interconvertible enzyme in response to increasing concentrations of e l (161, 181). These patterns differ with respect to their amplitude, signal amplification, and sensitivity to changes in e, concentration. In fact, three of these four regulatory patterns have been obseived in regulation of the mammalian pyruvate dehydrogenase cascade (183, 197). Furthermore, because the number of converter enzymes in multicyclic cascades is greater than in monocyclic cascades, it is possible to obtain an even greater number of unique regulatory patterns in response to positive and negative allosteric interactions. The fact that a minimum of two converter enzymes and one interconvertible enzyme is required to form a single interconversion cycle, and each enzyme can be a separate target for one or more allosteric effectors, cyclic cascades provide a high degree of flexibility for metabolic input. Through allosteric interactions with the cascade enzymes, fluctuations in the concentrations of numerous metabolites lead to automatic adjustments in the activities of the converter enzymes that determine the steady-state levels of fractional modification (specific activity) of the interconvertible enzymes. In essence, cyclic cascades serve as biological integrators that can sense changes in the concentrations of innumerable metabolites and modulate the activities of pertinent enzymes accordingly. 5 . Rate Amplification Kinetic analysis of multicyclic cascades reveals that the rate of covalent modification of the last interconvertible enzyme in the cascade is a multiplicative function of the rate constants of all the reactions that lead to the formation of the modified enzyme (8, 186). Therefore, following an initial lag period, cyclic cascades can function as rate amplifiers to generate an almost explosive increase in catalytic activity of the target interconvertible enzyme in response to stimuli. This rate amplification potential increases with the number of cycles in the cascade. The magnitude of the rate amplification is further enhanced if the multicyclic cascade involved possesses a pyramidal relationship with respect to the concentrations of its interconvertible enzymes. Employing reasonable esti-


33

mates for the rate constants, it has been demonstrated that multicyclic cascades are capable of generating large biochemical responses to primary stimuli in the millisecond time range. Moreover, if the converter and interconvertible enzymes are topographically positioned close to each other, an even greater rate of response is possible. Such topographic positioning does occur in the case of mammalian pyruvate dehydrogenase (198) and the enzymes of the glycogen cascade, which have been shown to be adsorbed to glycogen particles (199). Experimentally, Cori and his associates reported (200) that following electrical stimulation of frog sartorius muscle at 30"C, phosphorylase b is converted to phosphorylase a with a half-time of 700 msec. The fact that cyclic cascades can respond to stimuli in the millisecond time range, together with their capacity for signal amplification, indicates that they can be involved in the regulation of neurochemical processes (186).

6. ATP Flux As shown in Fig. 1, for each complete cycle of a phosphorylation-dephosphorylation cascade, one equivalent of ATP is consumed and one equivalent each of ADP and Pi is generated. Similarly, the energy-rich donor molecules for all other forms of covalent modification (see Table I) also are degraded continuously when the cyclic cascade is in operation. The rate of ATP hydrolysis is regulated by the parameters that control the fractional modification of the intercovertible enzyme. In the theoretical analysis (161, 181),the ATP concentration is assumed to be constant because, in vivo, the concentration of ATP is metabolically maintained at fairly constant levels which are several orders of magnitude greater than the concentrations of the enzymes involved in the cascades. This ATP flux is an essential feature of the cyclic cascade regulatory mechanism because it provides the free energy required to maintain the steady-state distribution between the modified forms of the interconvertible enzyme at metabolitespecified levels which are different from those specified by thermodynamic considerations (see Section IV).

111.

Experimental Verification of the Cyclic Cascade Model

To verify the properties of cyclic cascades described in the previous sections, the following systems have been investigated: (a) a simple in vitro phosphorylation-dephosphorylation cyclic cascade which was developed to study the validity of the theoretical predictions; (b) the bicyclic cascade of glutamine synthetase employing both purified proteins and permeabilized cells. The in v i m system consists of type I1 CAMP-dependent protein kinase and a

34

SHACTER. CHOCK, RHEE, AND STADTMAN

38-kDa, type 2A protein phosphatase (201) as converter enzymes, and a nanopeptide as interconvertible substrate (195). Both converter enzymes were purified to near homogeneity from bovine heart. The nanopeptide, Leu-Arg-ArgAla-Ser-Val-Ala-Gln-Leu, is homologous to the phosphorylation site of rat liver pyruvate kinase (202). The allosteric effectors used are CAMP, an activator of the kinase, and Pi, an inhibitor of the phosphatase. This model cascade is useful both as a tool for studying the mechanism whereby enzymes regulate each other and as a relatively simple system to aid our conceptualization of metabolic regulation through cyclic cascades. U$e results show that when the ATP concentration is maintained at a relatively 'constant level, a steady state is established for the fractional phosphorylation of the nanopeptide. Under this condition, ATP hydrolysis continues at a constant rate, which is a measure of the interconversion rate of the nanopeptide between its phosphorylated and dephosphorylatedforms. In addition, as predicted by the cyclic cascade analysis, this monocyclic cascade exhibits both the capacity for signal amplification and the capacity to generate a cooperative response to increasing effector concentrations. Due to signal amplification, only one-tenth of the CAMPconcentration required to half-activate the kinase is required to obtain a 50% phosphorylation of the nanopeptide. Because of an increase in sensitivity in this system, phosphorylation of the nanopeptide responds more sharply to increasing concentrations of CAMP than does the activation of the CAMP-dependent protein kinase. This apparent cooperativity derives from the fact that the catalytic subunit of the protein kinase forms a tight complex with the nanopeptide. Furthermore, in the presence of Pi, an inhibitor of the phosphatase, both the sensitivity and signal amplification were enhanced considerably (195). Experiments have been carried out also using purified proteins of the bicyclic glutamine synthetase cascade of E. coli (189, 203) and permeabilized E . coli cells (204). Despite complications resulting from the dual roles of the effectors glutamine and a-ketoglutarate in the adenylylation and deadenylylation reactions, and the fact that six of the converter enzyme-effector complexes are catalytically active, the data confirmed the predictions of the theoretical analysis. They illustrated that cyclic cascades can serve as a signal amplifier and can elicit a cooperative response to increasing effector concentrations. Similar results were obtained both in vitro and with permeabilized cells. In addition, comparative studies (204) show that cells in which both cycles of the bicyclic cascade are functioning possess a higher signal amplification and are more sensitive to changes in metabolite concentrations than those with only one cycle functioning. In essence, most of the predicted properties of cyclic cascades have been confirmed experimentally in v i m and in vivo through studies of the phosphorylation-dephosphorylation monocycle and the glutamine synthetase bicyclic cascade.


35

IV. Energy Consumption The steady states that develop in cyclic cascades are dynamic, energetic conditions in which the interconvertible proteins are being cycled constantly between modified and unmodified states. The capacity of a cyclic cascade system to maintain a steady state is dependent upon a constant supply of metabolic energy to drive the modification reactions, for in the absence of adequate donor molecules, the interconvertible protein would be converted completely to the unmodified form. Thus, the constant flux of metabolic energy through the cyclic cascade is the fuel required to maintain such an exquisite mechanism of cellular regulation. The question arises, then, as to how much energy is consumed by reversible covalent modification systems. Detailed quantitation of the ATP flux through a model monocyclic phosphorylation-dephosphorylation cascade (205) demonstrated that in the presence of a relatively constant amount of ATP, a number of general characteristics are observed, which can be summarized as follows: 1. The ATP turnover is directly proportional to the concentration of both converter enzymes in the system; i.e., the higher the concentration of both protein kinase and phosphatase, the higher the cycling rate and, hence, the higher the ATP flux. 2. Attainment of a specific steady-state level of phosphorylation is dependent upon the ratio of concentrations of protein kinase and phosphatase and is independent of their absolute concentrations. 3. The time required to reach a given steady state is inversely proportional to the concentrations of converter enzymes. 4. For a given concentration of protein kinase, phosphatase, and phosphorylatable protein, the rate of ATP consumption is directly proportional to the steady-state level of phosphorylation (determined by the effector concentrations for the converter enzymes); in other words, to maintain a protein in a highly phosphorylated, thermodynamically unstable state, proportionately more energy must be expended than to maintain it at a lower steady-state level of modification.

Note that these characteristics are relevant for all forms of reversible covalent modification. It seems likely that the cell must maintain a delicate balance between a requirement to reach a new steady state within the metabolic time frame while not expending an excessive amount of energy. In fact, the levels of converter enzymes in the cell are relatively high (in the p M range), and, not unexpectedly, steady-state levels of phosphorylation of specific proteins in vivo are reached within seconds after an extracellular stimulus (205). It was of interest, therefore,

36


to estimate what the actual ATP turnover in an in vivo cyclic cascade might be (205).To this end, all relevant experimental parameters were compiled for two well-characterized cellular cyclic cascade systems: phosphorylation of hepatic pyruvate kinase by CAMP-dependent protein kinase and phosphorylation of skeletal muscle glycogen phosphorylase by phosphorylase kinase following neural stimulation. The values were used to quantitate the ATP flux through each monocyclic cascade. It was estimated that each of these cyclic covalent modification systems consumes less than 0.02% of the total cellular energy flux. This analysis did not take into account the fact that multiple cascades are activated simultaneously following hormonal and neuronal stimuli, nor did it include the energy conservation mechanisms inherent in the cyclic cascade system (205). Nevertheless, even allowing for a large margin of error, this result suggests that cyclic cascade systems not only are exceptional in their regulatory potential, but they are highly energy efficient as well.

V. Covalent lnterconversion versus Simple Allosteric Control

Although the cyclic cascade model was derived for enzyme systems undergoing covalent interconversion, the reaction scheme shown in Fig. 1 theoretically does not require covalent modification. Hence, it is reasonable to question whether allosteric interactions alone between metabolites and enzymes can exhibit the properties of cyclic cascade systems, such as signal amplification, flexibility, and sensitivity with respect to increasing effector concentrations. It should be remembered that the cyclic cascade model utilizes both covalent modification of enzymes and allosteric control. To accomplish signal amplification by means of simple allosteric control, the following conditions would have to be met: (a) very tight binding between the allosteric effector and the target enzyme; and (b) a reaction that possesses catalytic properties such that one effector can activate more than one target enzyme molecule. Note that signal amplification in cyclic covalent modification cascades is achieved without a requirement for tight binding between effector and converter enzyme. Because the binding rate for the effector is limited by the diffusion rate, a slow off-rate for the enzyme-bound effector would be required to achieve tight binding. However, tight binding would reduce the temporal efficiency of the control process. Furthermore, in order to achieve a catalytic effect in a simple allosteric model, the effector would first have to bind to the target enzyme, induce an active conformation, and then dissociate from the active enzyme which would have to remain in the active conformation. Such a mechanism has been implicated in the past (206).However, to remain regulatable by the effector, the active enzyme would have to be able to relax back to its inactive form. This kind


37

of mechanism is thermodynamically unfavorable (207). Finally, without the presence of converter enzymes, the capacity for allosteric interactions would be reduced considerably. Nevertheless, the apparent cooperativity that provides the sensitivity observed in cyclic cascade systems can be accomplished by allosteric interaction alone, particularly if the enzyme involved contains multiple subunits. So, some of the advantages derived from cyclic cascade regulation cannot be achieved without invoking reversible covalent modification, while others can be accomplished but with less regulatory efficiency.

VI. Concluding Remarks The cyclic cascade model, derived mainly from data based upon detailed studies of glutamine synthetase, is applicable to all covalent interconvertible enzyme systems. It reveals many regulatory advantages such as signal amplification, rate amplification, sensitivity, and flexibility. This regulatory mechanism makes use of both covalent modification and allosteric interactions. By means of allosteric interactions with one or more enzymes, cyclic cascades can continuously monitor fluctuations in the concentrations of a multitude of metabolites and adjust the specific activities of the target enzymes according to biological requirements. Thus, they serve as biological integrators. Although a cyclic cascade modulates the specific activity of the interconvertible enzyme smoothly and continuously over a wide range of conditions, it can under extreme physiological situations serve as an ON-OFF switch to turn on or off an interconvertible enzyme. The energy for maintaining such an efficient regulatory mechanism is the consumption of ATP and other energy-rich donor molecules. However, the amount of ATP consumed is negligible compared to the total cellular ATP hydrolysis. In view of the unique properties of cyclic cascades, it is not surprising that a large number of key enzymes are regulated by this mechanism.

REFERENCES I . Krebs, E. G., and Beavo, J. A . (1979). Annu. Rev. Biochem. 48, 923-959. 2. Engstrom, L., Ekman, P., Humble, E., Ragnarsson, U., andZetterqvist, 0. (1984). “Methods in Enzymology,” Vol: 107, pp. 130-153. 3. Cooper, J . A , , Sefton, B. M . , and Hunter, T. (1983). “Methods in Enzymology,” Vol. 99, pp. 387-402. 4. Urushizaki, Y . , and Seifter, S. (1985). PNAS 82, 3091-3095. 5. MOSS,J . , and Vaughan, M. (1982). In “ADP-Ribosylation Reactions: Biology and Medicine” (0.Hayaishi and K . Ueda, eds.), pp. 637-645. Academic Press, New York. 6 . Ueda, K . , and Hayaishi, 0. (1984). “Methods in Enzymology,” Vol. 106, pp. 450-461. 7 . Bodley, J . W . , Van Ness, B. G . , and Howard, J . B. (1980). In “Novel ADP-Ribosylations of Regulatory Enzymes and Proteins” (M. Smulson and T. Sugimura, eds.), pp. 413-422. ElsevierlNorth-Holland, New York. 8. Chock, P. B . , Rhee, S. G . , and Stadtman, E. R. (1980). Annu. Rev. Biochem. 49, 813-843.

38


9. Bradley, M. K., Hudson, J., Villanueva, M. S . , and Livingston, D. M. (1984). PNAS 81, 6574-6578. 10. Paik, W. K. (1984). “Methods in Enzymology,” Vol. 106, pp. 265-268. 11. Paik, W. K., and Di Maria, P. (1984). “Methods in Enzymology,” Vol. 106, pp. 274-287. 12. Allfrey, V. G., Di Paola, E. A., and Sterner, R. (1984). “Methods in Enzymology,” Vol. 107, pp. 224-240. 13. Flavin, M., and Murofushi, H. (1984). “Methods in Enzymology,” Vol. 106, pp. 223-237. 14. Huttner, W. B. (1984). “Methods in Enzymology,” Vol. 107, pp. 200-223. 15. Sutherland, E. W., Jr., and Wosilait, W. D. (1955). Nature (London) 175, 169-170. 16. Fischer, E. H., and Krebs, E. G. (1958). JBC 231, 65-71. 17. Holland, R . , Hardie, D. G., Clegg, R. A., and Zammit, V. A. (1985). BJ 226, 139-145. 18. Kim, K.-H. (1983). Curr. Top. Cell. Regul. 22, 143-176. 19. Davis, C. G., Gordon, A. S . , and Diamond, I. (1982). PNAS 79, 3666-3670. 20. Huganir, R. L., Miles, K., and Greengard, P. (1984). PNAS 81, 6968-6972. 21. Walsh, M. P., Hinkins, S . , and Hartshorne, D. J. (1981). BBRC 102, 149-157. 22. Steinberg, R. A. (1980). PNAS 77, 910-914. 23. Field, F. J., Henning, B., and Mathur, S. N. (1984). BBA 802, 9-16. 24. Paxton, R., and Harris, R. A. (1982). JBC 257, 14433-14439. 25. Damuni, Z . , Caudwell, B., and Cohen, P. (1982). EJB 129, 57-65. 26. Wong, T. W., and Goldberg, A. R. (1984). JBC 259, 3127-3131. 27. Takahashi, T., Nakada, H., Okumura, T., Sawamura, T., and Tashiro, Y. (1985). BBRC 126, 1054-1060. 28. Jerebzoff, S . , and Jerebzoff-Quintin, S. (1984). FEBS Lett. 171, 67-71. 29. Ramakrishna, S . , Pucci, D. L., and Benjamin, W. B. (1981). JBC 256, 10213-10216. 30. Houston, B., and Nimmo. H. G . (1985). BBA 844, 193-199. 31. Sibley, D. R., Peters, J. R., Nambi, P., Caron, M. G., and Lefkowitz, R. J. (1984). JBC 259, 9742-9749. 32. Fujita, M., Taniguchi, N., Makita, A , , Ono, M., and Oikawa, K. (1985). BBRC 126, 818824. 33. Cook, K. G . , Bradford, A. P., Yeaman, S. J., Aitken, A,, Fearnley, I. M., and Walker, J. E. (1984). EJB 145, 587-591. 34. Michalak, M., Famulski, K., and Carafoli, G. (1984). JBC 259, 15540-15547. 35. Hartzell, H. C., and Glass, D. B. (1984). JBC 259, 15587-15596. 36. Hathaway, G. M., and Traugh, J. A. (1982). Curr. Top. Cell. Regul. 21, 101-127. 37. Erlichman, J., Rangel-Aldao, R., and Rosen, 0. M. (1983). “Methods in Enzymology,” Vol. 99, pp. 176-186. 38. Shoji, S . , Titani, K., Demaille, S. G., and Fischer, E. H. (1979). JBC 254, 6211-6214. 39. Lincoln, T. M., and Corbin, J. D. (1983). Adv. Cyclic Nucleotide Res. 15, 139-192. 40. Boyd, G. S., and Gorban, A. M. (1980). Mol. Aspecrs Cell. Regul. 1, 95-134. 41. Khoo, J. C., Mahoney, E. M., and Steinberg, D. (1981). JBC 256, 12659-12661. 42. Scallen, T. J., and Sanghvi, A. (1983). PNAS 80, 2477-2480. 43. Sharma, R. K., Wang, T. H., Wirch, E., and Wang, J. H. (1980). JBC 255, 5916-5923. 44. Sharma, R. K., and Wang, J. H. (1985). PNAS 82, 2603-2607. 45. Vilgrain, I . , Defaye, G., and Chambaz, E. M. (1984). BBRC 125, 554-561. 46. Hemming, H. C., Jr., Greengard, P., Tung, H. Y., and Cohen, P. (1984). Nature (London) 310, 503-505. 47. Durban, E., Roll, D., Beckner, G., and Busch, H. (1981). Cancer Res. 41, 537-545. 48. Ackerman, P., Glover, C. V. C., and Osheroff, N. (1985). !“AS 82, 3164-3168. 49. Cooper, J. A., Reiss, N., Schwartz, R. J., and Hunter, T. (1983). Narure (London) 302, 218223. 50. Cohen, S. (1983). “Methods in Enzymology,” Vol. 99, pp. 379-387. 51. Dekowski, S . A., Rybicki, A,, and Drickamer, K. (1983). JBC 258, 2750-2753.


39

52. Jagus, R., Crouch, D., Konieczny, A., and Safer, B. (1982). Curr. Top. Cell. Regul. 21, 3564. 53. Ochoa, S., de Haro, C., Siekierka, J . , and Grosfeld, H. (1981). Curr. Top. Cell. Regul. 18, 421-435. 54. Petryshyn, R., Levin, D. H., and London, 1. M. (1983). “Methods in Enzymology,” Vol. 99, pp. 346-362. 55. Kramer, G., and Hardesty, B. (1981). Curr. Top. Cell. Regul. 20, 185-203. 56. Lengyel, P. (1982). Annu. Rev. Biochem. 51, 251-282. 57. Traugh, J. A., and Lundak, T. S. (1978). BBRC 83, 379-384. 58. Tuhackova, Z . , Ullrichova, J . , and Hradec, J. (1985). EJB 146, 161-166. 59. Engstrom, L., Edlund, B., Ragnarsson, U., Dahlqvist-Edberg, U., and Humble, E. (1980). BBRC 96, 1503-1507. 60. Kadowaki, T., Nishida, E., Kasuga, M., Akiyama, T., Takaku, F., Ishikawa, M., Sakai, H . , Kathuria, S., and Fujita-Yamaguchi, Y. (1985). BBRC 127, 493-500. 61. Ekman, P., and Dahlqvist-Edberg, U. (1981). BBA 662, 265-270. 62. Claus, T. H., El-Maghrabi, M. R., Regen, D. M., Stewart, H. B., McGrane, M., Kountz, P. D., Nyfeler, F., Pilkis, J., and Pilkis, S. J . (1984). Curr. Top. Cell. Regul. 23, 57-86. 63. Wise, B. C . , Guidotti, A , , and Costa, E. (1984). Adv. Cyclic Nucleotide Protein Phosphorylation Res. 17, 511-519. 64. Singh, V. B., and Moudgil, V. K. (1985). JBC 260, 3684-3690. 65. Hemmings, B. A. (1978). JBC 253, 5255-5258. 66. Nimmo, G. A , , and Nimmo, H. G. (1984). BJ 224, 101-108. 67. Krebs, E. G. (1981). Curr. Top. Cell. Regul. 18, 401-419. 68. Roach, P. J . (1981). Curr. Top. Cell Regul. 20, 45-105. 69. Woodgett, J. R., and Cohen, P. (1984). EJB 143, 267-272. 70. Akatsuka, A , , Singh, T. J., Nakabayashi, H., Lin, M. C., and Huang, K. P. (1985). JBC 260, 3239-3242. 71. Detre, J . A , , Naim, A. C . , Aswad, D. W., and Greengard, P. (1984). J . Neurosci. 4, 28432849. 72. Zwiller, J . , Revel, M. O., and Malviya, A. N. (1985). JBC 260, 1350-1353. 73. Taylor, S. S. (1982). JBC 257, 6056-6063. 74. Walton, G . M., Spiess, J., and Gill, G. N. (1982). JBC 257, 4661-4668. 75. Langan, T. A., Zeilig, C., and Leichtling, B. (1981). In “Protein Phosphorylation” (0.M. Rosen and E. G . Krebs, eds.), pp. 1039-1052. Cold Spring Harbor Lab., Cold Spring Harbor, New York. 76. Feurstein, N., Monos, D. S., and Cooper, H. L. (1985). BBRC 126, 206-213. 77. Belfrage, P., Fredrikson, G . , Olsson, H., and Stralfors, P. (1984). Adv. Cyclic Nucleoride Protein Phosphorylation Res. 17, 35 1-359. 78. Steinberg, D. (1976). Adv. Cyclic Nucleotide Res. 7, 157-198. 79. Ingebritsen, T. S., and Gibson, D. M. (1980). Mol Aspects Cell. Regul. 1, 63-93. 80. Beg, Z. H., Stonik, J . A , , and Brewer, H. B., Jr. (1985). JBC 260, 1682-1687. 81. Ingebritsen, T. S. (1983). Biochem. Soc. Trans. 11, 644-646. 82. Beg, Z. H., Stonik, J. A., and Brewer, H. B., Jr. (1984). BBRC 119, 488-498. 83. Swamp, G . , Dasgupta, J . D . , and Garbers, D. L. (1984). Adv. Enzyme Regul. 22, 267-288. 84. Cobb, M. H., and Rosen, 0. M. (1984). BBA 738, 1-8. 85. Kasuga, M., Karlsson, F. A , , and Kahn, C. R. (1982). Science 215, 185-187. 86. Zick, Y., Whittaker, J . , and Roth, J. (1983). JBC 258, 3431-3434. 87. Shackelford, D. A., and Trowbridge, I . S. (1984). JBC 259, 11706-1 1712. 88. Leonard, W. J., Depper, J. M., Kronke, M . , Robb, R. J., Waldmann, T. A , , and Greene, W. C. (1985). JBC 260, 1872-1880. 89. Garland, D., and Nimmo, H. G. (1984). FEBS Lett. 165, 259-264. 90. Gibbs, P. E., Zouzias, D. C., and Freedberg, 1. M. (1985). BBA 824, 247-255.

40


91. Theurkauf, W. G., and Vallee, R. B. (1982). JBC 257, 3284-3290. 92. Turner, R. S., Chou, C. H., Mazzei, G. J., Dembure, P., and Kue, J. F. (1984). J . Neurochem. 43, 1257-1264. 93. Hasilik, A., Pohlmann, R., Olsen, R. L., and von Figura, K. (1984). EMBO J. 3,2671-2676. 94. Albanesi, J. P., Fujisaki, H., and Korn, E. D. (1984). JBC 259, 14184-14189. 95. England, P. J. (1984). J. Mol. Cell. Curdiol. 16, 591-595. 96. Walsh, M. P., Hinkins, S., Dabrowska, R., and Hartshorne, D. J . (1983). “Methods in Enzymology,” Vol. 99, pp. 279-288. 97. Westwood, S. A,, Hudlicka, 0.. and Perry, S. V. (1984). BJ 218, 841-847. 98. Adelstein, R. S. (1983). J . Clin. Invest. 72, 1863-1866. 99. Costa, M. R., Casnellie, J. E., and Catterall, W. A. (1982). JBC 257, 7918-7921. 100. Mardh, S. (1983). Curr. Top. Membr. Trunsp. 19, 999-1004. 101. Atmar, V. J., and Kuehn, G. D. (1983). “Methods in Enzymology,” Vol. 99, pp. 366-372. 102. Golf, S. W., and Graef, V. (1984). J. Clin. Chem. Clin. Biochem. 22, 705-709. 103. Donlon, J., and Kaufman, S. (1978). JBC 253, 6657-6659. 104. Wretborn, M., Humble, E., Ragnarsson, U., and Engstrom, L. (1980). BBRC 93, 403-408. 105. Soling, H.-D., and Brand, I. A. (1981). Curr. Top. CeN. Regul. 20, 107-138. 106. Le Peuch, C. J., Haiech, J . , and Demaille, J . G. (1979). Biochemistry 18, 5150-5157. 107. Villalba, M., Varela, I., Mdrida, I., Pajares, M. A., Martinez del Pozo, A , , and Mato, J. M. (1985). BBA 847, 273-279. 108. Cohen, P. (1982). Nature (London) 296, 613-620. 109. Resink, T. S., Hemmings, B. A., Tung, H. Y. L., and Cohen, P. (1983). EJB 133,455-461. 110. Jurgenson, S., Shacter, E., Huang, C. Y., Chock, P. B., Yang, S.-D. Vandenheede, J. R., and Merlevede, W. (1984). JBC 259, 5864-5870. 1 11. Villa-Moruzzi, E., Ballou. L. M., and Fischer, E. H. (1984). JBC 259, 5857-5863. 112. Singh, T. J . , Akatsuka, A., and Huang, K. P. (1984). JBC 259, 12857-12864. 113. Katbh, N., and Kubo, S. (1983). BBA 760, 61-68. 114. Stiles, C. D. (1983). Cell 33, 653-655. 115. Jacob, S. T., and Rose, K . M. (1984). Adv. Enzyme Regul. 22, 485-497. 116. Stetler, D. A., Seidel, B. L., and Jacob, S. T. (1984). JBC 259, 14481-14485. 117. Schaffhausen, B. S., and Benjamin, T. L. (1979). Cell 18, 935-946. 118. Oetting, W. S., Tuazon, P. T., Traugh, J. A., and Walker, A. M. (1986). JBC 261, 16491652. 119. Qi,D. F., Turner, R. S., and Kuo, J. F. (1984). J. Neurochem. 42, 458-465. 120. Reed, L. J . (1981). Curr. Top. Cell. Regul. 18, 95-106. 121. Pettit, F. H., Yeaman, S. J., and Reed, L. J. (1983). “Methods in Enzymology,” Vol. 99, pp. 331-336. 122. Engstrom, L. (1978). Curr. Top. Cell. Regul. 13, 29-51. 123. Burnell, J. N., and Hatch, M. D. (1985). TrendsBiochem. Sci. 10, 288-291. 124. Lee, S. G., Miceli, M. V., Jungmann, R. A., and Hung, P. P. (1975). PNAS 72,2945-2949. 125. Shichi, H., Somers, R. L., and Yamamoto, K. (1983). “Methods in Enzymology,” Vol. 99, pp. 362-366. 126. Cobb, M. H., and Rosen, 0. M. (1983). JBC 258, 12472-12481. 127. Trevillyan, J. M., Perisic, O., Traugh, J. A., and Byus, C. V. (1985). JBCMO, 3041-3044. 128. Rose, K. M., Duceman, B. W., Stetler, D., and Jacob, S. T. (1983). Adv. EnzymeRegul. 21, 307-320. 129. Nestler, E. J., Walaas, I., and Greengard, P. (1984). Science 225, 1357-1364. 130. Yamamoto, H., Fukunaga, K., Goto, S . , Tanaka, E., and Miyamoto, E. (1985). J. Neurochem. 44, 759-768. 131. Perry, S . V. (1979). Biochem. Soc. Trans. 7, 593-617. 132. Card. D. L.. and Kirschncr. M. W. ,1985) I Cell Bio! 100 769-774.

2.

CYCLIC CASCADES AND METABOLIC REGULATION

41

133. Albert, K. A,, Helmer-Matyjek, E., Nairn, A. C., Muller, T. H., Haycock, J. W., Greene, L. A,, Goldstein, M., and Greengard, P. (1984). PNAS 81, 7713-7717. 134. Cahill, A. L., and Perlman, R. L. (1984). PNAS 81, 7243-7247. 135. Werth, D. K., Niedel, J. E., and Pastan, 1. (1983). JBC 258, 11423- 11426. 136. Kun, E. Romaschin, A. D., Claisdell, R. J., and Jackowski, G. (1981). In “Metabolic Interconversion of Enzymes” (H. Holzer, ed.), pp. 280-293. Springer-Verlag, Berlin and

New York. 137. Moss, J., and Vaughan, M. (1984). “Methods in Enzymology,” Vol. 106, pp. 411-418. 138. Nomura, H., Tanigawa, Y., Kitamura, A , , Kawakami, K., and Shimoyama, M. (1981). BBRC 98, 806-814. 139. Creissen, D., and Shall, S. (1982). Nature 296, 271-272. 140. Yoshihara, K., Itaya, A,, Tanaka, Y., Ohashi, Y., Ito, K., Teraoka, H., Tsukada, K., Matsukage, A,, and Kamiya, T. (1985). EBRC 128, 61-67. 141. Ferro, A. M., and Olivera, B. M. (1984). JBC 259, 547-554. 142. Van Ness, B. G., Howard, J. B., and Bodley, J. W. (1980). JBC 255, 717-720. 143. Moss, J., Watkins, P. A , , Stanley, S. S . , Purnell, M. R., and Kidwell, W. R. (1984). JBC 259, 5100-5104. 144. Poirier, G. G., Niedergang, C., Champagne, M., Mazen, A., and Mandel, P. (1982). EJB 127, 437-442. 145. Ueda, K., Ogata, N., Kawaichi, M., Inada, S., and Hayaishi, 0. (1982). Curr. Top. Cell. Regul. 21, 175-187. 146. Tanaka, Y., Yoshihara, K., Ohashi, Y., Itaya, A,, Nakano, T., Ito, K., and Kamiya, T. (1985). Anal. Biochem. 145, 137-143. 147. Pope, M. R., Murrell, S. A,, and Ludden, P. W. (1985). PNAS82, 3173-3177. 148. Tsuchiya, M., Tanigawa, Y., Ushiroyama, T., Matsunra, R., and Shimoyama, M. (1985). EJB 147, 33-40. 149. Wong, N. C. W., Poirier, G. G., and Dixon, G. H. (1977). EJB 77, 11-21. 150. Goff, C. G. (1984). “Methods in Enzymology,” Vol. 106, pp. 418-429. 151. Goldman, N., Brown, M., and Khoury, G. (1981). Cell 24, 567-572. 152. Abood, M. E., Hurley, J. B., Pappone, M. C., Bourne, H. R., and Stryer, L. (1982). JBC 257, 10540-10543. 153. Niles, E. G., and Westhead, E. W. (1973). Biochemistry 12, 1723-1729. 154. Kim, S. (1984). “Methods in Enzymology, Vol. 106, pp. 295-309. 155. Stock, I . B., Clarke, S., and Koshland, D. E., Jr. (1984). “Methods in Enzymology,” Vol. 106, pp. 310-321. 156. Huszar, G. (1984). “Methods in Enzymology,” Vol. 106, pp. 287-295. 157. K q , D. R. (1983). JBC 258, 12745-12748. 158. Liu, M.-C., and Lipmann, F. (1984). PNAS 81, 3695-3698. 159. Baeuerle, P. A,, and Huttner, W. B. (1984). EMBO J. 3, 2209-2215. 160. Garland, D., and Russell, P. (1985). PNAS 82, 653-657. 161. Stadtman, E. R., and Chock, P. B. (1978). Curr. Top. Cell. Regul. 13, 53-95. 162. Johnston, M. I., and Torrence, P. F. (1984). In “Interferon, 3: Mechanisms of Production and Action” (R. M. Friedman, ed.), pp. 189-298. Elsevier, Amsterdam. 163. Sugimura, T., and Miwa, M. (1983). Carcinogenesis (London) 4, 1503-1506. 164. Sefton, B. M., and Hunter, T. (1984). Adv. Cyclic Nuckotide Protein Phosphorylation Res. 18, 195-226. 165. Collett, M. S . , and Erikson, R. L. (1978). PNAS 75, 2021-2024. 166. Levinson, A. D., Oppermann, H., Levintow, L., Varmus, H. E.. and Bishop, J. M. (1978). Cell 15, 561-572. 167. Tanigawa, Y.,Tsuchiya, M., Imai, Y., and Shimoyama, M. (1983). BBRC 113, 135-141. 168 Sols. A . (1981). Curr. Top. CeN. Renu/. 19, 77-101.

42


Parker, R. A., Miller, S. J., and Gibson, D. M. (1984). BBRC 125, 629-635. Bergstrom, G., Ekman, P., Humble, E., and Engsuom, L. (1978). BEA 532, 259-267. Hemmings, B. A. (1980). FEBS Lett. 122, 297-302. Muller, D., and Holzer, H. (1981). BBRC 103, 926-933. Toyo-oka, T. (1982). BBRC 107, 44-50. Hershko, A., and Ciechanover, A. (1982). Annu. Rev. Biochern. 51, 335-364. MacFarlane, R. G. (1964). Nature (London) 202, 498-499. Neurath, H., and Walsh, K. A. (1976). In “Proteolysis and Physiological Regulation” (E. W. Ribbons and K. Brew, eds.), pp. 29-40. Academic Press, New York. 177. Davie, E. W., and Fujikawa, K. (1975). Annu. Rev. Biochem. 44, 798-829. 178. Miiller-Eberhard, H. J . (1975). Annu. Rev. Biochem. 44, 697-724. 179. Kingdon, H. S . , Shapiro, B. M., and Stadtman, E. R. (1967). PNAS 58, 1703-1710. 180. Wulff, K., Mecke, D., and Holzer, H. (1967). BBRC 28, 740-745. 181. Stadtman, E. R., and Chock, P. B. (1977). PNAS 74, 2761-2765. 182. Brown, M. S . , Segal, A., and Stadtman, E. R. (1974). ABB 161, 319-327. 183. Pettit, F. H., Pelley, J. W., and Reed, L. 1. (1975). BBRC 65, 575-582. 184. Chock, P. B., and Stadtman, E. R. (1977). PNAS 74, 2766-2770. 185. Chock, P. B., and Stadtman, E. R. (1979). In “Modulation of Protein Function” (D. E. Atkinson and C. F. Fox, eds.), pp. 185-202. Academic Press, New York. 186. Stadtman, E. R., and Chock, P. B. (1979). I n “The Neurosciences: Fourth Study Program” (F. 0. Schmitt, ed.), pp. 801-817. MIT Press, Cambridge, Massachusetts. 187. Garcia, E., and Rhee, S. G. (1983). JBC 258, 2246-2253. 188. Caban, C. E., and Ginsburg, A. (1976). Biochemistry 15, 1569-1580. 189. Rhee, S. G., Park, R., Chock, P. B., and Stadtman, E. R. (1978). PNAS 75, 3138-3142. 190. La Porte, D. C., and Koshland, D. E., Jr. (1982). Nature (London) 300, 458-460. 191. El-Maghrahi, M. R., Claus, T. H., Pilkis, J., Fox, E., and Pilkis, S. J. (1982). JBC 257, 7603-7607. 192. Van Schaftingen, E., Davies, D. R., and Hers, H. G. (1982). EJE 142, 143-149. 193. Jurgensen, S. R., Chock, P. B., Taylor, S. S., Vandenheede, J. R., and Merlevede, W. (1985). FP 44, 1052. 194. Shacter-Noiman, E., Chock, P. B., and Stadtman, E. R. (1983). Philos. Trans. R . SOC. London 302, 157-166. 195. Shacter, E., Chock, P. B., and Stadtman, E. R. (1984). JBC 259, 12252-12259. 196. Goldbetter, A., and Koshland, D. E., Jr. (1981). PNAS 78, 6840-6844. 197. Hucho, F., Randell, D. D., Roche, T. E., Burgett, M. W., Pelley, J. W., and Reed, L. J. (1972). ABB 151, 328-340. 198. Reed, L. J. (1969). Curr. Top. Cell. Regul. 1, 233-251. 199. Meyer, F., Heilmeyer, L. M. G., Jr., Haschke, R. H., and Fischer, E. H. (1970). JBC 245, 6642-6648. 200. Danforth, W. H., Helmreich, E., and Cori, C. F. (1962). PNAS 48, 1191-1 199. 201. Shacter-Noiman, E., and Chock, P. B. (1983). JBC 258, 4214-4219. 202. Titanji, V. P. K., Ragnarsson, U . , Humble, E., and Zetterqvist, 0. (1980). JBC 255, 1133911343. 203. Rhee, S. G., Chock. P. B., and Stadtman, E. R. (1985). In “The Enzymology of PostTranslational Modification of Proteins” (R. Freedman, ed.), Vol. 2, pp. 273-297. Academic Press, New York. 204. Mura, U., Chock, P. B., and Stadtman, E. R. (1981). JBC 256, 13022-13029. 205. Shacter, E., Chock, P. B., and Stadtman, E. R. (1984). JBC 259, 12260-12264. 206. Hatfield, G. W., and Bums, R. 0. (1970). Science 167, 75-76. 207. Astumian, D., and Chock, P. B. (1985). Unpublished results. 169. 170. 171. 172. 173. 174. 175. 176.

Cyclic NucleotideDependent Protein Kinases STEPHEN J. BEEBE JACKIE D. CORBIN Howard Hughes Medical Institute Department of Molecular Physiology and Biophysics Vanderbilt University Nashville, Tennessee 37232

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Purification ............................. A. Cyclic AMP-Dependent Protein Kinase . . . B. Cyclic GMP-Dependent Protein Kinase ........................... 111. Characterization and Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Cyclic AMP-Dependent Protein Kinases B. Cyclic GMP-Dependent Protein Kinase ........................... C. Evolutionary Relationships among Cyclic AMP-Binding Proteins and Protein Kinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Mechanism of Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Mechanism of Regulatory Subunit Action B. Mechanism of Catalytic Subunit Action .......................... V. Biological Role of Protein Kinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Criteria to Establish Biological Role . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Distribution of Isozymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Selective Activation of Cyclic AMP-Dependent Protein Kinase Isozymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Use of Cyclic Nucleotide Analogs in Intact Cells . . . . . . . . . . . . . . . . . . . E. The Role of Cyclic AMP and the Cyclic AMP-Dependent Protein Kinase Isozymes in the Cell Cycle, Proliferation, and Differentiation F. Variations in the Regulatory Subunit-Catalytic Subunit Ratio

48

49 49 59 60 64

64 67 69 69 71 73 80

85

93

43 THE ENZYMES,Vol. XVII Copyright 0 1986 by Academic Press, Inc. All rights of reproduction in any form reserved

44

STEPHEN J. BEEBE AND JACKIE D. CORBIN G . Regulation of the Amount of Certain Proteins by Cyclic AMPDependent Protein Kinases ..................................... References .....................................................

1.

96 100

Introduction

Since the first report of a protein kinase in liver by Burnett and Kennedy (I) and the classical work on interconvertible forms of enzymes involved in glycogen metabolism, as well as investigations of protein kinases reported by Rabinowitz (2), phosphorylation and dephosphorylation reactions have been the most intensely studied mechanisms of posttranslational modification. Many different protein kinases have since been identified and studied, but the cyclic nucleotide-dependent protein kinases have probably received the most attention. The primary impetus for research in this area was the Nobel Prize-winning work of Sutherland and co-workers who developed the concept of cAMP as an intracellular second messenger of hormone action (3, 4), and the discovery of the CAMP-dependentprotein kinase (5). It was eventually recognized that in mammalian cells the CAMP-dependent protein kinase is the major, if not the only, intracellular receptor for CAMP.The activation-inactivation reaction is indicated by the following equation (6):

+

R2Cz 4 CAMP (inactive)

R ~ ( c A M P )+~ 2C (active)

In the holoenzyme form (R2C2), the catalytic subunit (C) is inhibited by the regulatory subunit (R). The CAMP-dependent protein kinase is represented by two different major types of isozymes which are operationally defined by the salt gradient elution behavior from DEAE-cellulose. Type I elutes as NaCl concentrations less than 0.1 M and type I1 elutes at concentrations greater than 0.1 M (7). Both holoenzymes are tetramers composed of dimeric regulatory subunits with four CAMP-bindingsites (two sites/monomeric chain) (6-9) and two monomeric catalytic subunits. When cAMP binds to the regulatory subunit of protein kinase the equilibrium shifts to the right and the catalytic subunit is released from the inhibition imposed by the regulatory subunit. It is the free, active catalytic subunit that mediates protein phosphorylation, which is established to be the primary, if not the only, mechanism of cAMP action in mammals. When cAMP is hydrolyzed by cAMP phosphodiesterases (see Chapter 2), the equilibrium shifts back to the left, the catalytic subunit reassociates with the regulatory subunit and phosphotransferase activity is terminated (IO-I5). In order for the unmodified form of the protein substrate to be regenerated, a phosphoprotein phosphatase catalyzes a dephosphorylation reaction (see Chapter 8). The steadystate, phosphorylation-dephosphorylation equilibrium between an active and

3. CYCLIC NUCLEOTIDE-DEPENDENT PROTEIN KINASES

45

inactive form of a protein is dynamically regulated by protein kinases, phosphoprotein phosphatases, and their respective effectors. This constitutes a cyclic cascade system and can provide the cell with an efficient and sensitive control mechanism that has potential for amplification and cooperativity of response (16-20, 20u). The ubiquity of the cyclic nucleotide protein kinases in nature and the central role of cyclic nucleotides in eukaryotic metabolic regulation makes the involvement of cyclic nucleotides as effectors of this system of particular interest and importance. The cGMP-dependent protein kinase, which has received less attention, may also play a role in the phosphorylation-dephosphorylation control of cellular events. It has a different mechanism of activation-inactivation as indicated by the following equation (21): E2

+ 4 cGMP (inactive)

E2 . c G M P ~

(active)

The cGMP kinase is a dimer with each monomeric chain containing both a regulatory, cGMP-binding component and a catalytic component (21-24). There are four cGMP-binding sites per enzyme dimer or two sites in each of the two regulatory components. When cGMP binds to the regulatory domain of the enzyme, the equilibrium shifts to the right and the active catalytic domain carries out the phosphorylation reactions. No separation of subunits occurs (see Ref. 29) as is the case for activation of the CAMP-dependent protein kinase, where the regulatory and catalytic subunits physically separate. When cGMP dissociates from the active form of the enzyme and is hydrolyzed by cGMP phospbdiesterase, the equilibrium shifts back to the left and the inactive conformation of the enzyme is reestablished. The reader is referred to several reviews (5, 10-15, 25-40) and the monograph edited by Rosen and Krebs (41), which are related to cyclic nucleotidedependent protein kinases and their role in phosphorylation-dephosphorylation mechanisms. The aim of this chapter is to review the characteristics and functions of cyclic nucleotide-dependent protein kinases. A brief review of methods for purifying the enzyme is presented. The characteristics of the enzymes are described as they relate to functional aspects of the isozymes, to evolutionary relationships and homologies among protein kinases and other cyclic nucleotide binding proteins, and to the mechanisms of catalytic and regulatory subunit action. The regulation of enzyme activity by cyclic nucleotides and other effectors is reviewed and the role the enzymes play in cellular function is assessed. The role of these kinases in the short-term control of cellular function through the regulation of enzyme activities and the long-term control of processes such as transcription, protein induction and cell growth, and differentiation are highlighted. In addition, several methods used for deducing the biological role of the kinases are compared and critiqued.

46

STEPHEN J. BEEBE AND JACKIE D. CORBlN

II. Purification A.

CYCLICAMP-DEPENDENT PROTEINKINASE

1. Regulatory Subunit Before the development of affinity chromatography the regulatory subunit of the CAMP-dependent protein kinase could be obtained by purification of the respective holoenzymes followed by subunit dissociation. Since then the regulatory subunit can be more easily purified to homogeneity by affinity chromatography using immobilized cAMP analogs ( 4 2 4 5 ) . Both type I and type I1 regulatory subunits can be purified using this approach, but it is often advantageous to use different affinity ligands for each (46). The regulatory subunits can be eluted from affinity columns using 8 M urea, followed by urea removal and renaturation. Alternatively, they can be specifically eluted by CAMP. The urea elution method allows the recovery of a relatively CAMP-free regulatory subunit but it usually has altered properties compared to the subunit prepared by cAMP elution (47). Even though elution with cAMP alleviates some anomalies, removal of bound cAMP from the regulatory subunit without deleterious effects is very difficult (6). The type I regulatory subunit binds very effectively to N6-(2-aminoethy1)amino-CAMP-Sepharose(44) and is readily purified to homogeneity by specific cAMP elution (44,46).The type I subunit is not readily eluted from 8-(6arninohexy1)amino-CAMP-Sepharose using cAMP but can be eluted using 8 M urea (46). This subunit has also been purified using a N6-(6-aminohexyl)-cAMP derivative attached to Sepharose (45). The subunit is not eluted from this ligand using 2.5 M salt or 0.5 mMcAMP but is eluted with 0.5 mM cAMP in the presence of a low concentration of catalytic subunit. The type I1 regulatory subunit has been prepared to homogeneity by Corbin et al. (6, 46) by using a cAMP elution from 8-(6-aminohexyl)amino-cAMPSepharose. The type I1 subunit has also been purified using a cAMP or cGMP (0.1 mM) elution from N6-aminoethyl-cAMP-Sepharose(48). Since cGMP binds weakly and is thus more easily removed, it has the advantage of allowing the preparation of a relatively cyclic nucleotide-free regulatory subunit. In order to obtain a nondenatured, CAMP-free regulatory subunit, Seville and Holbrook (49) used a method where cGMP is exchanged for cAMP while the subunit is bound to DEAE-cellulose. The subunit is then eluted with salt and the weakly bound cGMP is removed by washing and dialysis. In general, the type I1 regulatory subunit is eluted from cyclic nucleotide affinity columns at lower cAMP concentrations than is the type I subunit. For example, the elution of type I1 subunit from 8-(6-aminohexyl)amino-cAMP-Sepharose is carried out with 10 mM cAMP (6)but elution of type I requires urea denaturation (46). While the type I1 subunit

.


47

is eluted from N6-aminoethyl-CAMP-Sepharoseat 0.1 mM cAMP or 0.1 mM cGMP, the type 1 subunit is eluted with 30 mM cAMP (48).

2. Catalytic Subunit The catalytic subunit of the CAMP-dependent protein kinase has been purified from a large number of tissues including bovine heart; rabbit, rat, and porcine skeletal muscle; bovine and rabbit liver [see Ref. (38) for review]; rabbit kidney and porcine stomach mucosa (50);rat adipose tissue (51); and porcine kidney (52). The most efficient purification procedures are based on the facts that the holoenzyme and regulatory subunit have different ion exchange properties from the catalytic subunit, and that the subunits specifically dissociate in the presence of cAMP or cAMP analogs (50).Two different approaches have been used to purify the catalytic subunit to homogeneity. A partially purified preparation of type I and/or type I1 holoenzyme is treated with cAMP before chromatography on carboxymethylcellulosewhich binds the catalytic subunit but not the regulatory subunit (53,54). Alternatively the holoenzyme can be treated with cAMP while it is bound to DEAE-cellulose which does not retain the dissociated catalytic subunit (50).Subsequent purification or concentration of catalytic subunit may be necessary on hydroxylapatite (50, 5 3 , carboxymethylcellulose (53), Sephadex G-100 (52, 54), or Blue Dextran (52). 3 . Holoenzymes The holoenzymes are usually obtained in pure form by first purifying the regulatory and catalytic subunits by the procedures previously described. These subunits are then combined. It is not clear, however, whether or not in all cases the holoenzyme produced by this recombination resembles in all respects the native holoenzymes. Although the methods are more difficult, it may be necessary for some studies to purify the holoenzymes as such. The partial purification of the CAMP-dependent protein kinase was first reported by Walsh et al. (56). Since then the holoenzymes have been purified to homogeneity in a number of laboratories. The type I isozyme has been purified from rabbit (53,57-59) and porcine skeletal muscle (60, 61) and the type I1 holoenzyme from bovine heart (62-64). The rabbit skeletal muscle type I and the bovine heart type I1 are often taken as the prototype isozymes. These tissues are ideal for purifying the respective enzymes since they contain predominately one isozyme. Conventional purification steps have been used to purify the enzymes. The isozymes are first separated on DEAE-cellulose based on their different salt elution (7). C,-Aminoalkyl-agarose chromatography has also been used to separate the isozymes (65).The procedures for most homogeneous preparations of either holoenzyme have successfully employed DEAE-cellulose chromatography, ammonium sulfate fractionation, alumina C, chromatography, and gel filtration techniques. Beavo et al. (53, 54) and Hofmann et al. (57) used negative chro-

48

STEPHEN J. BEEBE AND JACKIE D. CORBIN

matography by batch adsorption to remove some contaminating proteins on carboxymethylcellulose. Hydroxylapatite has also been used (58, 59, 61-64), but Rubin et al. (63) found that in their procedure, rechromatography on DEAEcellulose could substitute for the hydroxylapatite step. Hofmann et al. (57) used histone IIA-Sepharose to purify the bovine heart type I1 kinase, and Taylor et al. (60, 61) used 8-(6-aminohexyl)amino-ATP-Sepharose and 3-aminopyridineNAD -Sepharose affinity chromatography in addition to isoelectric focusing to purify porcine muscle type I holoenzyme to homogeneity. Hydrophobic chromatography on hexyl-Sepharose (58) or phenyl-Sepharose (64) has also been used as very effective purification steps for type I and type I1 isozymes, respectively. Cobb and Corbin (64) have successfully used Bio-Rad 1EX545-DEAE and Bio-sil TSK-250 high-performance liquid chromatography in addition to conventional procedures to purify the type II holoenzyme from bovine heart to homogeneity. These homogeneous preparations are generally purified 10003000-fold depending on the tissue and the isozyme. +

B.

KINASE CYCLICGMP-DEPENDENT PROTEIN

The cyclic GMP-dependent protein kinase has been purified to homogeneity from soluble fractions of bovine lung by Lincoln et al. (66), Gill et al. (22), Corbin and Doskeland ( 2 4 , and MacKenzie (24) and from bovine heart by Flockerzi et al. (67). A unique form of the enzyme has also been highly purified from intestinal brushborder membranes by de Jonge (68). The purification procedures generally utilize DEAE-cellulose chromatography, ammonium sulfate precipitation, and cyclic nucleotide affinity Chromatography. All of the cyclic nucleotide-dependent protein kinases bind to DEAE-cellulose, and this is a convenient early step in the purification procedure. The cGMP-dependent protein kinase from bovine lung binds more tightly to DEAEcellulose than does the type I CAMP-dependent protein kinase and less tightly than does the type I1 isozyme of the cAMP protein kinase (22, 66), and is separated from the latter enzymes using NaCl(Z2, 60) or ammonium sulfate (22, 66, 67, 69, 70). Corbin and Doskeland (21) also used a procedure through the DEAE-cellulose step designed for purification of guanylate cyclase (71). Triethanolamine (pH 7.5) elutes the cGMP-dependent protein kinase slightly before it elutes the cyclase on this column. At least three different immobilized cyclic nucleotide analogs have been used to obtain pure preparations of the cGMP kinase. Sepharose-bound 8-(2-aminoethyl)thio-cGMP was first used by Lincoln et al. (66) and 8-(2-aminoethyl)aminocAMP by Gill et al. (22) to purify the enzyme to homogeneity. Corbin and Doskeland (21) and MacKenzie (24) used 8-(6-aminohexyl)amino-cAMP-Sepharoseto purify the cGMP kinase in their studies. When N6-[(6-aminohexyl)carbamoylmethyl]-cAMPwas used to purify the enzyme, it was not homogeneous (24). Since the CAMP-dependent protein


49

kinases also bind to these affinity resins, preliminary purification on DEAEcellulose is important. Additionally, specific elution of the enzyme from the affinity columns with cGMP allows for a high degree of purification. If trace contamination by regulatory subunit is still present at this stage, it can be removed by sucrose gradient centrifugation ( 2 1 ) . The purification of the cGMPdependent protein kinase from intestinal brushborder membranes requires a modified procedure (68). Following the extraction of the enzyme from the membranes with detergent and high salt concentration, it is further purified on 8-(2aminoethy1)-amino-CAMP-Sepharose.Except for the presence of 0.1% Triton and high salt concentration during all steps of affinity chromatography, the procedure is similar to the ones for purification of the lung enzyme. Several factors can be manipulated to maximize or minimize binding of these cyclic nucleotide protein kinases to affinity supports (48, 72).In addition to the relative affinities for binding, other factors such as steric hindrance, density of substitution, and availability of ligand can be of potential importance.

111.

Characterization and Physical Properties

A.

CYCLICAMP-DEPENDENT PROTEIN KINASES

There are two major classes of CAMP-dependent protein kinase isozymes designated type I and type I1 (58). Each isozyme has a tetrameric structure consisting of two monomeric catalytic subunits and a dimeric regulatory subunit (65-68). The catalytic subunits from both isozymes are indistinguishable by a large number of criteria. They have similar chromatographic, chemical, physical, immunological, and catalytic properties, and can reassociate with both type I and type I1 regulatory subunits ( 1 1 , 12, 14, 35, 38, 55). However, several investigators have found two or three different froms of catalytic subunit with different isoelectric points from both type I and type I1 isozymes (55, 73-78). Sugden et al. identified at least three forms of the bovine liver catalytic subunit with pl values of 6.72,7.04, and 7.35 (55). Yamamura et al. (74-76) found that the rabbit skeletal muscle type I and the rat liver type I1 catalytic subunits each contained two forms with isoelectric points of 7.4 and 8.2. These forms have similar heat stability, K,,,for ATP, and rate of phosphorylation of several proteins; they phosphorylate the same serine and threonine residues in histone, protamine, glycogen synthase, and phosphorylase (74- 76).The significance of these different isoelectric forms of the catalytic subunit is not clearly understood. It has been discovered that there are at least two bovine genes coding for the catalytic subunit. One of these gene codes for a protein of 351 amino acids that is 98% homologous with the bovine heart catalytic subunit (78a).The second gene codes for a protein closely related to the bovine heart catalytic subunit. The

50

STEPHEN 1. BEEBE AND JACKIE D. CORBIN

nucleotide sequences of the two genes are 85-93% homologous, with differences clustered in the sequences coding for the amino terminal portion of the molecules, believed to be the ATP binding site (79), in the carboxy terminal region of the protein, and in the 3’-untranslatedportion of the gene (S. McKnight and R. Maurer, personal communication). It is presently unclear whether the products of these two genes are related to the different isoelectric forms of the catalytic subunit discussed above. However, this is possible since the two genes code for proteins which would be expected to have different isoelectric points and, consequently, different tryptic peptides. It is also possible that the second gene could be related to the “mute” catalytic subunit like the one isolated and characterized from rat skeletal muscle by Reed et al. (786).The “mute” subunit is released from the regulatory subunit by cAMP but must be activated by a heatand acid-stable modulator. Some evidence suggests that these two genes may have different tissue distributions. For example, the mRNA for the second gene appears to be more abundant in brain (S. McKnight and R. Mauer, personal communications). Although there is presently no evidence to correlate the association of either of these catalytic subunit gene products with one or the other regulatory subunits, it is known that the type IIB (85) or neural type I1 regulatory subunit (125, 126) is also abundant in the brain. These recent developments suggest a potential for greater specificity of protein kinase action, and firm conclusions regarding possible functional differences should be forthcoming. Future work will undoubtedly focus on the expression of these two genes and possibly others, so that their protein products can be characterized. The amino acid sequence of the catalytic subunit from type I1 bovine heart has been determined to consist of 350 residues (79),giving it a molecular weight of 40,862, including a myristyl amino terminal blocking group and phosphates at threonine-197 and serine-338. This molecular weight is in fairly good agreement with a molecular weight of 39,000-42,000 determined by SDS-gel electrophoresis, sedimentation-equilibrium centrifugation, amino acid analysis, or calculated from the Stokes radius (2.7 nm) and sedimentation coefficient = 3.6) (38).The protein has a frictional ratio of 1.2 and an axial ratio of 4.5, indicating a globular, symmetrical shape (38, 55). At present it is known that there is at least one gene for the type I regulatory subunit and at least two genes for the type I1 regulatory subunit. The type I subunit has been cloned from bovine testes (78a),and one of the type I1 subunits has been cloned from rat ovary. The gene for the type I regulatory subunit codes for a protein with the same amino acid sequence as the rabbit skeletal muscle type I. The DNA for the rat ovary type I1 subunit codes for a M, = 52,000 protein, which is distinct from the bovine heart type I1 subunit ( 7 9 4 . These two type I1 subunits are homologous, have similar amino-terminal and carboxylterminal amino acid sequences, and have two duplicated sequences which are presumably the cAMP binding domains. However, these two duplicated se-


51

quences and the sequences surrounding the autophosphorylationsite are different between the two proteins. Sequences around the autophosphorylationsite of the rat ovary type I1 regulatory subunit are also different from sequences in the same area of the rat heart regulatory subunit (J. S. Richards, personal communication). Further developments in this field will undoubtedly strengthen our understanding of these proteins and their functions. The type I and type I1 isozymes are homologous proteins which have the same tetramenc structure, similar mechanisms of CAMPactivation (see Section II,C), and two different intrasubunit cyclic nucleotide-binding sites on the regulatory subunit (6,59).Although the sequence homologies of the regulatory subunits are strongest in the CAMP-binding domains, the sequences of other domains are less well conserved (see Section 11,C) and several features of the regulatory subunits or their respective holoenzymes can be used to differentiate them. These include physical, immunological, and kinetic differences (10, 11, 14, 38, 57). The type I1 subunit has an axial ratio of 12 and a frictional rateo of 1.6 (82),indicating an oblong asymmetric structure. It consists of 43% a helices, 23-30% P-strands, and has 23 p-turns (81). The type I regulatory subunit has an axial ratio of 8.5 and a frictional ratio of 1.47 (86).The molecular weights have been determined from amino acid sequence to be 42,804 for the bovine skeletal muscle type I regulatory subunit (379 residues) (80) and 45,004 for the bovine heart type I1 subunit (400 residues) (81). These values are slightly lower than the molecular weights determined by sedimentation equilibrium centrifugation (15,82)or nondenaturing gel electrophoresis (34, 62, 83) according to the method of Hedrick and Smith (84), but considerably lower than the apparent molecular weights determined from SDS-gel electrophoresis [M, = 49,000 for the type I from porcine, bovine, and rabbit skeletal muscle and 56,OOO-58,000 for the bovine heart type I1 subunit Refs. (14, 57)]. Robinson-Steineret al. (85) showed that the apparent molecular weight of the type I1 regulatory subunits determined by SDSgel electrophoresis differ significantly among different species and tissues. In addition, the extent of migration differs for some forms of type I1 regulatory subunit depending on the presence of phosphate in the autophosphorylation site (33,34,85).The type I regulatory subunit does not undergo autophosphorylation and shows similar but slightly less pronounced discrepancies in molecular weights when determined by different methods (14, 59). The explanation for the error in apparent molecular weight for the type I1 regulatory subunit determined by SDS-gel electrophoresis is not certain. It is possible that the protein structure may not be completely denatured by SDS and therefore does not migrate strictly according to molecular weight. If this is correct, the presence or absence of phosphate on the autophosphorylation site affects the binding of SDS more to some than to other forms of type I1 regulatory subunit (85). The physical properties of the holoenzymes have been reviewed and conveniently tabulated by Nimmo and Cohen (14) and by Carlson et al. (38). The

52

STEPHEN J. BEEBE AND JACKIE D . CORBIN

molecular weights of the holoenzymes are now known from amino acid sequences to be 167,332 for type I (79, 80) and 171,732 for type I1 (79, 81). Estimates of molecular weights from sedimentation equilibrium centrifugation and nondenaturing gel electrophoresis (62, 83) are essentially the same as these values. The isolation of the CAMP-dependentprotein kinases by anion-exchangechromatography, using DEAE cellulose, has been used as a standard procedure to define, separate, and quantitate the type I and type II isozymes (7, 35). The different elution behaviors are predicted from their respective isoelectric points (PI) and differences in amino acid composition. The type I subunit has a higher pl(5.45-5.57) (87, 88) and elutes at lower salt concentrations than does the type I1 subunit, which has a pl of 5.34-5.40. In addition the type I1 isozyme has a higher acidic amino acid content (6). Using a standard, low-ionic-strength phosphate buffer, the type I and type I1 isozymes elute from DEAE cellulose at less than and greater than 100 mM NaCl, respectively. The type I and type I1 holoenzymes have distinct antigenic determinants (8992). When the holoenzymes or the regulatory subunits are used as antigens, antibodies are produced that recognize the individual regulatory subunits (89, 92). Specific antibodies to the catalytic subunit are generated only when the free subunit is used as the antigen, and the immunological properties appear to be similar among the subunits from distinct tissues and species (15, 54, 55, 93). Although antisera are isozyme specific, they lack absolute species specificity. For example, Fleischer er al. (89) generated antibodies against the bovine heart type I1 isozyme that cross-reacted identically with the type I1 isozymes from other bovine tissues, but reacted in a nonparallel manner with the type I1 isozymes from rat tissues and procine heart. Weldon et al. (94) prepared monoclonal antibodies against the type I1 bovine heart regulatory subunit that had an antigenic site localized in a region near the dimerization domain, the autophosphorylation site, and the CAMP-binding site 2. This was a conserved sequence in the bovine and porcine heart isozymes and was recognized with similar affinities in both species by one monoclonal antibody. The type I and type I1 isozymes are also different in some of their kinetic properties and these have been used to distinguish the isozymes when mixtures are present. The type I holoenzyme is more readily dissociated than the type I1 holoenzymes in the presence of histone and high salt concentration (7, 35, 95). The presence of MgATP prevents the dissociation of the type I holoenzyme under these conditions (7, 57) probably by inhibition of CAMPbinding (57). The type I1 holoenzyme reassociates rapidly in the absence of salt but slowly in the presence of salt, while the type I holoenzyme associates slowly in the presence or absence of salt (7, 35). The presence of phosphate in the autophosphorylation site of the type I1 regulatory subunit [serine-95of the primary sequence (81, 96)] increases the dissociation constant for the regulatory and catalytic subunit com-


53

plex and slows the rate of reassociation (9, 34, 97, 98). Although the physiological significance of the autophosphorylation reaction is not clear, it has been proposed that the type I1 holoenzyme exists in vivo primarily in the phosphorylated form (99), which is apparently favored for dissociation (9). Recently, Scott and Mumby (100)demonstrated that both phosphorylated and dephosphorylated forms exist in intact trachael smooth muscle and that the relative amounts are changed by intracellular CAMP. The type I1 regulatory subunit is also phosphorylated in vitro at serines-44 and -47 by glycogen synthase kinase 3 (101) and at serines-74 and -76 by casein kinase I1 (101, 102). These latter two serines are also phosphorylated in vivo (102). The physiological significance of these phosphorylations, if any, has not been established. Other kinetic differences have been revealed through the use of cAMP analogs. Although the cAMP binding domains have been highly conserved (see Section II,C), differences in cyclic nucleotide binding to the type I and type I1 isozymes have been found. Although both the 3' and 5' oxygens in the ribose portion of cAMP are important for activation of both isozymes, use of analogs substituted with sulfur at these positions suggests that differences exist between the two isozymes in one or both of the binding sites that recognize the 4' position of the ribose ring and the 3',5' cyclic phosphate (103). In addition, C8-aminoalkyl-CAMP analogs preferentially activate the type I isozyme and 2-phenyl- 1, N6-etheno-CAMP, the only significant one of over 100 analogs tested, preferentially activates the type I1 isozyme (104). More striking differences between the binding sites of cyclic nucleotide-dependent isozymes are evident when both CAMP-binding sites are considered (Fig. 1). For the CAMP-dependent protein kinase isozymes as well as the cGMPdependent protein kinase, site-1 (site B) has a slower cAMP dissociation rate (59, 86, 105-107) and a relative selectivity for cAMP analogs modified at the C8 carbon of the base (C8 analogs) (86, 105, 107), while site-2 (site A) has a faster dissociation rate and a relative selectivity for analogs modified at the C6 position (C6 analogs) with the CAMP-dependent protein kinase isozymes. This site has a relative selectivity for analogs modified at the C1 position (C1 analogs) for cGMP analogs with the cGMP-dependent protein kinase (108). [3H]cIMP (site-2 selective), 8-a~ido-[~~P]cAMP (site- 1 selective), and [3H]cAMP-bindingexperiments establish that binding of cyclic nucleotides at either site stimulates binding at the other site for both type I and type I1 protein kinase isozymes (107, 109, 110). This positive cooperativity in cyclic nucleotide binding is reflected by a positive cooperativity of protein kinase activation (111). The cAMP and cGMP kinases have Hill coefficients of approximately 1.6 for their respective activators (12, 85, 112). In addition, a synergism of protein kinase activation for each isozyme occurs using a combination of a site 1 and site 2 selective analog (111, 113). An important point in this regard is that the analog combinations that cause synergism of protein kinase activation differ for the isozymes. The type I iso-

54


Lu

C6-ANALOGS

4 TYPE

C 8 AMNO-ANALOGSC-

I cAMP

KINASE

f$

C 6 -ANAL OGS

TYPE IL cAMP KINASE C 8 TMIO-ANALOGS

OR

8- PIPERIDIN0- C A W

NI-ANALOGS

cGMP KINASE C8- ANALOGS

FIG. I . Proposed structural homologies between CAMP-dependent and cGMP-dependent protein kinases. The cyclic nucleotide analog selectivities for the two intrasubunit binding sites are indicated.

zyme generally shows better relative selectivity at site I for C8 amino analogs and the type I1 isozyme generally shows a better relative selectivity at site 1 for C8 thio analogs, while both isozymes show selectivity at site 2 for C6 analogs. Consequently, a combination of a C6 and a C8 amino analog causes a synergism of type I but not type I1 activation while a combination of a C6 and a C8 thio analog causes a synergism of type I1 but not type I activation (1 11, 113). An unusual analog, 8-piperidino-cAMP, has been characterized (104). This analog is selective for site 1 on the type I1 isozyme but selective for site 2 on the type I isozyme. Therefore, 8-piperidino-CAMPis used with C8 amino analogs as a type I directed-analog pair or used with C6 analogs as a type I1 directed-analog pair. [3H]cGMP-bindingto the cGMP-dependent protein kinase is stimulated by analogs that are selective for site 2, but not by those selective for site 1 (108); This enzyme is synergistically activated by a combination of a C1 and a C8 analog. Interestingly, cGMP itself also exhibits synergism with a C1 analog (108). By using cAMP analogs that are highly selective for one site on the regulatory subunit of one isozyme, analog combinations can be chosen that maximize the


55

synergism of the response of a single isozyme in a tissue that contains both isozymes. Ogreid et al. (104) have used a quantitative in vitro approach which should predict the best analog combination to use for in vivo experiments so that only a single isozyme is activated. The C8 amino analogs are most highly selective for site 1 on the type I isozyme, but analogs that are optimally selective for other sites on the respective isozymes are unavailable. Analog pairs that have been used in intact tissues (113) may not measure the optimal synergism potential for a given isozyme. However, when appropriate site-selective analogs are used in intact tissues at relatively low concentrations so that the protein kinase(s) is activated only slightly above its basal state (5-15% of maximum), a slight elevation of the physiological response occurs. The synergism observed with an appropriate type I- or type 11-directed analog pair can then be attributed to a single isozyme with a minimum synergistic activation of the other isozyme (114). Regardless of whether optimal pairs of analogs are used or not, optimal synergism will be observed only when the protein kinase and the cell responses are slightly activated above the basal state with single analogs. While the classification of isozymes into major type I and I1 classes is useful, the development of more advanced techniques to prove the isozyme structure indicates that at least the type II kinase is represented by a continuum of microheterogeneous forms. The DEAE-cellulose elution behavior of several type I1 isozymes is different (7, 35, 112). Malkinson et al. (115) found that an apparent murine adipose tissue type I elutes at a higher salt concentration and contaminates the type I1 isozyme. Toru-Delbauffe et al. (116) reported a similar result with the type I1 isozyme from rat thyroid. Robinson-Steiner et al. (85)found that a second fractionation on DEAE-cellulose is sometimes required to completely resolve the two isozymes. This is especially the case when the type I to type I1 ratio is high. However, several potential artifacts can result in an incorrect isozyme identification from DEAE-cellulose analysis alone. The free regulatory subunits of both isozymes elute at higher salt concentrations than do the holoenzymes (35, 117), and type I regulatory subunit can contaminate the type I1 holoenzyme. This can be a complication when the holoenzymes are artifactually activated by the homogenization procedure. It is also difficult to rule out that during chromatography free catalytic subunit reassociates with free regulatory subunit and modifies the protein kinase elution behavior. Limited proteolysis of the two regulatory subunit isozymes produces similar fragments with apparent molecular weights of 30,000 to 40,000 (118), and the DEAE-cellulose elution behavior of partially proteolyzed type I and type I1 isozymes may be less well resolved (119). It has been demonstrated that limited trypsinization of the type I1 bovine heart holoenzyme produces a CAMP-dependent dimeric enzyme, containing a proteolyzed regulatory subunit of M, = 45,000-48,000 on denaturing gels and a single intact catalytic subunit (120, 120a) (see Fig. 4 and Section III,A). A similar molecule was found in aged preparations of type I1 protein kinase from

56

STEPHEN J . BEEBE AND JACKIE D. CORBIN

rat liver (121) and adipose tissue (S. J. Beebe and J. D. Corbin, unpublished). Since both the dimeric enzyme from bovine heart (120) and the native adipose tissue holoenzyme (35, 112) elute at a relatively low salt concentration on DEAE-cellulose and each has a regulatory subunit with an apparent molecular weight by SDS-gel electrophoresis of M, = 48,000 (120) and M, = 5 1,000 (85, 112), respectively, isozyme classification can be erroneous. Since the type 11, but not the type I, regulatory subunit is autophosphorylated (57, 122, 123), a simple method is available to aid the identification of the type I1 isozyme. The type I isozyme can be identified since MgATP inhibits cAMP binding to this holoenzyme but not to the type I1 holoenzyme (57, 124). When the existence of the two main isozyme types of the CAMP-dependent protein kinase was first described (35, 112), it was pointed out that there is more than one class of type I1 in the same animal species. It was found that rat adipose tissue and rat heart type I1 elute at different NaCl concentrations from DEAEcellulose and that mixtures of the enzymes from the two tissues can be separated by chromatography on these columns. In spite of these physical differences, they were found to exhibit similar kinetic properties. From immunological studies of type I1 regulatory subunit, Erlichman et al. (125, 126) described neural and nonneural subclasses. Corbin et al. (85, 112) suggested that this classification may not include all type I1 subforms. By comparing mobility upon gel electrophoresis, Stokes radii measurements, and the effect of autophosphorylation and proteolysis on the type I1 regulatory subunit from different tissues and species, several different forms could be resolved (85).An operational classification separates type I1 regulatory subunits into those that shift mobility on SDS gels after autophosphorylation (type HA) and those that do not (type IIB) (85). These forms are further distinguished by apparent molecular weights. An examination of the distribution of types IIA and IIB in different species of heart tissue revealed a definite pattern (Fig. 2) (85).Type IIB is present in hearts of rodents, lagomorphs, and primates: while type IIA is present in hearts of carnivores and ungulates. These two broad groups of species diverged from each other about 70 million years ago. That type IIB is a more recent evolutionary development is suggested by the finding of type IIA in chicken heart. More than one form is also found in different tissues of the same species. Bovine lung contains equal amounts of M, = 56,000 type IIA (like bovine heart) and M, = 52,000 type IIB. Bovine brain contains a small amount (-15%) of this same type IIA and a predominant M, = 52,000 type IIB. The same forms are also found in different species. Both rat brain and bovine brain contain type IIB. The rat adipose tissue (M,= 51,000 type IIB) and the bovine heart (M,= 56,000 type IIA) holoenzymes differ in several other properties, including Stokes radius, calculated molecular weight, and frictional ratio (85, 112). Even though the adipose tissue enzyme is quite similar to other type I1 forms in the kinetics of cAMP action, the

57


HUMAN CHI MPA N 2 EE RHESUS ATELES

SLOW L O R I S

WHALE H I PPOPOTAMU s

GUANACO CAMEL

2 EBRA

Y

DONKEY I 0

I

I

I

1

1

40

80

120

160

200

MILLIONS OF YEARS B P

FIG. 2. Distribution of types IIA and B in hearts of various species. Shaded symbols represent type IIA and cross-hatched symbols represent type IIB. Different geometric symbols indicate molecular weight differences as determined by SDS-polyacrylamide gels: circle, M, = 56,000 (dephospho-); parallelogram, M, = 54,000 (dephospho-); ellipse, M, = 52,000; rectangle, M, = 51,000. The evolutionary tree is from Goodman (1034.

58


CAMP-binding sites are different from them based on the kinetics of activation by certain CAMP analogs. Several analogs modified at the N6-position of the adenine ring have higher apparent K, values for protein kinase activation for the adipose tissue enzyme than for the bovine heart and several other heart isozymes (112). The M, = 51,OOO type IIB enzymes from bovine brain and monkey heart, like that from adipose tissue, show similar but less striking differences compared to the bovine heart and other heart isozymes (S. J. Beebe and J. D. Corbin, unpublished). Jahnsen er af. (127) purified and characterized three isoforms (M, = 54,000, 52,000, and 5 1 ,000) of the type I1 regulatory subunits from rat ovarian granulosa cells using immunological, electrophoretic, photoaffinity labeling, and phosphorylation criteria. Antiserum against rat heart type I1 regulatory subunit recognizes the M, = 54,000 form while antibovine heart type I1 regulatory subunit serum recognizes the other two forms. The M, = 51,000 and 52,000 forms, unlike the M, = 54,000 form, are regulated by hormone in preovulatory follicles and do not readily change mobility on SDS-gels following phosphorylation. In addition, the M, = 54,000 form has a distinct peptide map. The authors suggest that the hormone-regulated forms are distinct gene products. Other apparent type I1 forms have been described. Schwartz and Rubin (128) have identified two distinct forms of M, = 54,000 and 52,000 from Friend erythroleukemic cells using monoclonal antibodies and peptide map analysis. Further distinctions between subclasses of the type I1 isozyme are indicated by differences in the amino acid composition, two dimensional tryptic peptide maps, and the peptide containing the autophosphorylation site between porcine brain and skeletal and cardiac muscles (129). Weldon et af. (130) have reported that the type I1 regulatory subunit from bovine brain binds only 2 mol of CAMP/ mol subunit, although the two different classes of binding sites are present. This is in contrast to bovine heart and other regulatory subunit isozymes which bind 4 mol cAMP/mol subunit. The explanation for the lower total CAMP-binding to the brain subunit requires further study. Furthermore, two different monoclonal antibodies with antigenic determinants at the NH,-terminal third of the heart regulatory subunit class react very poorly with the brain subunit. This information, in conjunction with comparative amino acid sequence analysis of this region suggest that these two forms of type I1 regulatory subunit are unique gene products (130). In summary, the demonstration of microheterogeneous forms of the type I1 isozyme tends to blur the distinction between isozyme forms. Consequently a careful examination using a series of tests is required to positively identify an isozyme. Although the number of isozymes studied in detail is small, the kinetic differences between the isozymes is most reliable. In addition, no type I isozymes have been reported to undergo autophosphorylation.


59

B. CYCLICGMP-DEPENDENT PROTEINKINASE The soluble cGMP-dependent protein kinase from bovine lung is a dimer composed of two identical monomeric chains each containing 670 amino acid residues (131). The monomer has a molecular weight of 76,33 1 determined from the amino acid sequence. Each chain contains a regulatory domain with two homologous cGMP-binding domains and a catalytic domain carboxyl-terminal to the binding domains (131). The dimer has a molecular weight of 152,662. This is in fairly good agreement with the molecular weight of 150,000-162,000 determined from SDS-gel electrophoresis (22, 66) or calculated from the sedimentation coefficient (6.7s-7.8s) (66, 132), and the Stokes radius (5.0-5.2 nm) (66, 132). The molecular weight of the cGMP kinase is therefore only slightly less than that of the cAMP kinase and considerably higher than that of the catalytically active component of the cAMP kinase. The cGMP-dependent kinase has an isoelectric point of 5.4 and is probably somewhat less asymmetric than the cAMP kinase since both the axial ratio of 7.4 and the frictional ratio of 1.4-1.5 are slightly less (132). Like the cAMP kinase, the cGMP kinase is autophosphorylated (70). Although this occurs in the presence of either cGMP or cAMP (25, 133), autophosphorylation with cAMP is distinctly different from that with cGMP. First, 2-6 mol phosphate per mol subunit is incorporated, instead of 0.75 mol/mol with cGMP (134). Stimulation of phosphate incorporation into seines-50 and -72 and theonines-58 and -84 (134) occurs with the occupancy of only one of the two intrasubunit binding sites (135). The most rapidly phosphorylated residue, theonine-58, is the major site phosphorylated in the presence of cGMP (134). Second, autophosphorylation in the presence of CAMP, but not cGMP, causes a 10-fold reduction in the concentration of cAMP required for half-maximal activation of phosphotransferase activity (133). Furthennore, following maximal autophosphorylation (total of -4 mol phosphate/mol subunit) the dissociation rate of cGMP from site 1, measured with an excess of cold cGMP, is decreased approximately 1O-fold. Autophosphorylation therefore primarily affects the binding at site 1 and elimates cooperative binding at this site (136). Other properties of the soluble cGMP-dependent protein kinase have been more completely reviewed by others (25-29) and the effects of cyclic nucleotide analogs are discussed in conjunction with the CAMP-dependent protein kinase in Section II,A. De Jonge (68) has characterized an intestinal, brush border membrane-specific cGMP-dependent protein kinase (type 11) that is distinct from the enzyme characterized in lung, heart, and smooth muscle (type I). In contrast to the soluble enzyme, the brush border cGMP kinase has an apparent molecular weight of 86,000 and is anchored to the membrane or to the contractile core of the microvilli by a 15,000-dalton fragment containing a site which is preferentially phos-

60


phorylated in situ in a cGMP-dependent manner. An apparent M, = 7 1,000 form, which contains the cGMP-binding and catalytic domains, is generated from the apparent M, = 86,000 form by proteolysis. The brushborder cGMP kinase is further differentiated from the soluble enzyme by isoelectric point (7.5) and phosphopeptide pattern following limited proteolysis. However, the brushborder cGMP kinase is immunologically similar to the soluble form and also undergoes autophosphorylation in the absence of cGMP.

C, EVOLUTIONARY RELATIONSHIPS AMONG CYCLIC AMP-BINDINGPROTEINS AND PROTEIN KINASES The first proposal of homology between protein kinases was based on similarities in physical and kinetic properties, and in amino acid composition, between CAMP- and cGMP-dependent protein kinases (23, 29). A model for structural homology between the two enzymes was also proposed (25), and it was suggested that the two intrasubunit cyclic nucleotide-binding sites on each enzyme were evolved by contiguous gene duplication (10,21,25,86).Structural models, slightly modified from the original, are shown in Fig. 3. Conclusive proof of a common progenitor for these kinases was the subsequent finding of a high degree of amino acid sequence homology in both the regulatory and in the catalytic domains (131). In the case of the regulatory component, the homology occurs not only between the two intrasubunit cyclic nucleotide-binding sites in the carboxy terminal two-thirds of this sequence, suggesting a tandem gene duplication, but also between the corresponding binding sites of the two different cyclic nucleotide-dependent protein kinases. There is much less apparent homology between these kinases in the amino terminal one-third of the primary sequence, which contains the dimerization domain. As expected, the regulatory subunits of the two main isozymic forms of CAMP-dependent protein kinase, types I and 11, also exhibit strong homology. The finding of several microheterogeneous forms of type I1 regulatory subunit, which can generally be classified into types IIA and IIB subclasses (85), and the finding that the primary amino acid sequence of a short segment of at least one type IIB form is very different from that of type IIA (85), are additional lines of evidence that the entire primary sequence of the regulatory subunit has not been well conserved. The regulatory and catalytic components of cyclic nucleotide-dependent protein kinases have been independently derived in the course of evolution since there is no apparent homology between them. Each of them is a member of one or the other of two distinct families, the CAMP-binding proteins and protein kinases, as illustrated in Fig. 3. All higher eucaryotes examined so far contain the regulatory components of cyclic nucleotide-dependent protein kinases. The enzymes in fungi (137-143, slime mold (146), and insects (147-149) appear to have the same basic structure as the mammalian enzymes. Vardanis has reported

PROTEIN KINASES

cAMP BINDING PROTEINS

S€RlNE- THR€WlNE PROTEIN KINASK3

w

PHOSPHORYLAS€ KINASE MYOSIN LIGHT CHAIN KINASE A

\\

PROCAR VOWS

4fl

OTHERS

€UCARVOT€S COOH

VIRUS ENCODE0 PROTEIN KINASES

CAMP - DEPENDEN PROJ€lN KINAS€

r

EGF RECEPTOb INSUL IN RECEPTOR OTHER GROWTH FACTOR RECEPTtS

OTHER CYCLIC NUCLEOTIDE BINDING PROTEINS DICTYOSTELIUM CAMP RECEPTOR TRYPANOSOMA cAMP BINDING PROTEIN c GMP-BINDING PHOSPHODIESTERASES 1

cGMP -D€fENDENT PROJElN KINASE

FIG.3. Possible evolutionq relationships among cyclic nucleotide binding proteins and protein kinases.

62


the existence in insects of a protein kinase that lacks specificity with regard to cAMP and cGMP (150), and of a CAMP-dependent protein kinase in nematodes, which does not dissociate into subunits during activation (151). There is no consistent evidence of cyclic nucleotide-dependent protein kinases in procaryotes, but these organisms contain a CAMP-bindingprotein (152, 153) named catabolite gene activator protein (CAP). This protein serves a similar function in E. coli as the regulatory subunit serves in the mammalian liver: when sugar availability declines, cAMP is elevated, binding of cAMP to the respective protein occurs and this leads to increased sugar availability. In mammals sugar is made available by increased glycogenolysis and gluconeogenesis; in bacteria sugar is made available by the switching on of catabolic gene operons for proteins involved in various sugar transport and utilization. Thus, CAP contains, in addition to a CAMP-bindingdomain, a domain at the carboxy terminus (indicated in black in the model of Fig. 3) which binds to DNA and regulates gene expression. This domain is known to have structural and sequence homologies with the DNA-binding protein of bacterial and viral repressors (154,155). The CAMP-binding domain of CAP is strongly homologous with the CAMP-binding domains of cyclic nucleotide-dependent protein kinases (156). From both X-ray crystallographic analysis of CAMP-bound CAP and from sequence homology between CAP and regulatory subunit, certain predictions of cAMP binding to cyclic nucleotide-dependent protein kinases have been made (156). It has been suggested that each CAMP-binding site contains a p-roll structure and perhaps an a-helix. If the regulatory components of the kinases are like CAP they could contain a deep pocket for cAMP binding between the p-roll and a-helix. Each site presumably also contains an essential arginine for binding the ribose phosphate moeity of CAMP. The CAP protein is an asymmetric dimer, the monomeric components presumably containing identical amino acid sequences (155). Two molecules of cAMP bind to the dimer, and each molecule binds to both monomers. The binding of cAMP to CAP exhibits cooperativity (157). In several respects the two CAMPbinding sites of the dimeric CAP resemble the two intrasubunit CAMP-binding sites of the regulatory components of the protein kinases. This suggests that one possible evolutionary precursor of the CAMP-binding sites of each protein is a dimeric, asymmetric precursor as illustrated in Fig. 3. CAP could have evolved from it by acquiring a DNA-binding domain through gene fusion. The regulatory subunit could have evolved from it by acquiring at least a catalytic subunitbinding domain and a dimerization domain at the amino terminus. In addition, by gene duplication and fusion, two CAMP-binding sites on the same protein chain could have been produced. These two sites might then have similar relative emplacements as the two sites on the dimeric CAP. If this theory is correct, then one might predict that the two cooperative intrasubunit cyclic nucleotide-binding sites of the regulatory component of protein kinase could be close to each other, and that each cyclic nucleotide binds to components of both binding domains.

3 . CYCLIC NUCLEOTIDE-DEPENDENT PROTEIN KINASES

63

Other cyclic nucleotide-binding proteins which are not components of a protein kinase have been described. These should be examined to see if they are related in evolution to the CAP proteins and to the cyclic nucleotide-binding sites of cyclic nucleotide-dependent protein kinases. In the cellular slime mold Dictyosrelium discoideum, cAMP acts as a chemotactic agent to induce cell aggregation during exhaustion of food supplies (158). This effect is mediated by a cell surface receptor which has physical and kinetic properties different from either CAP or the cyclic nucleotide-dependent protein kinases (159, 160). In species of Trypanosoma, a parasitic lower eucaryote, cAMP is believed to be involved in the regulation of cell growth and differentiation. These organisms are low or completely devoid in cyclic nucleotide-dependent protein kinase activity, but they contain a CAMP-binding protein of M, = 62,000 which differs from the kinase or CAP protein in several respects (161). Another well-known cyclic nucleotide-binding protein is the family of phosphodiesterases which possess a cGMP-specific-binding site in addition to a catalytic site (162-164). As seen in Fig. 3, the protein kinase family of proteins includes the catalytic component of CAMP- and cGMP-dependent protein kinases. This family can conveniently be divided into serine-threonine-specific protein kinases and tyrosine-specific protein kinases (see Chapter 6). As is the case for cyclic nucleotide-dependent protein kinases, two Ca2 -calmodulin regulated protein kinases, phosphorylase kinase and myosin light chain kinase, catalyze phosphorylation of serine or threonine in protein substrates. These kinases are also known to be closely related to the cyclic nucleotide kinases by amino acid sequence homology (165, 166). More distantly related to the catalytic component of cyclic nucleotide-dependent protein kinases is the group of tyrosinespecific protein kinases (167). A lysyl residue that binds the ATP analogue affinity label [(fluorosulfonyl)-benzolladenosineis apparently conserved in all of the protein kinases and could represent a part of the active site that interacts with the terminal phosphate of ATP (162, 164). It is not unreasonable to expect that other protein kinases, which have similar physical and kinetic properties but which have not yet been examined at the protein chemical level of those shown in Fig. 3, will be shown to be in this same family of proteins related to each other in evolution. Speculation is in order for the mechanism by which “marriage” of ancestral protein kinase catalytic proteins and inhibitory proteins such as the regulatory subunit occurred. One possibility is that originally the regulatory protein was simply a protein kinase substrate. Through mutational alteration, this protein could have developed such a strong affinity that it prevented, in a competitive manner, the phosphorylation of other protein substrates. Support for this theory is derived from the fact that the regulatory components of the cyclic nucleotidedependent kinases contain either a phosphorylation site or phosphorylation analog site for the catalytic component of the enzymes. Furthermore, the phosphorylation site is believed to be located in the catalytic component inhibitory +

64


region of the regulatory component (6, see Ref. 174), even though the phosphorylation site is known not to be the only component of the inhibitory domain of the regulatory subunit (4). Finally, it is known that several different protein kinases catalyze autophosphorylation (10). Whether or not the phosphorylation site of any of these enzymes is in an inhibitory domain has not been established.

IV. Mechanism of Action A. MECHANISM OF REGULATORY SUBUNIT ACTION The precise mechanism by which the regulatory subunit inhibits the catalytic subunit is not certain, although some progress has been made in understanding this phenomenon. In higher organisms all of the isozymes that have been characterized have a tetrameric structure (R2C2),but the functional unit of these forms appears to be a dimer (RC). This proposal is based on the observation that proteolytically modified type I1 holoenzyme (R’C) from either bovine liver (121), rabbit skeletal muscle (120),or bovine heart (120, 120~) behaves as a dimer (Fig. 4) on sucrose density gradients and gel filtration. The modified enzymes of muscle and heart have been studied in detail and it has been determined that the site of proteolysis, using either endogenous proteases or added trypsin, is in the amino terminal onefourth of the molecule (120, 1 2 0 ~ )It. has been further established that the carboxy terminal monomeric fragment, which contains the two CAMP-binding sites and the autophosphorylation site, retains virtually unaltered ability to inhibit the catalytic subunit. Thus, even though the native dimeric structure of regulatory subunit has been highly conserved, it is apparently not required for the inhibitory action. Perhaps all forms of holoenzyme do not contain a dimeric regulatory subunit. The inhibitory domain(s) of the type I1 regulatory subunit must be contained, at least in part, around the autophosphorylation site (residue 95) of the primary sequence (total residues = 400). This conclusion is based on the observation that modifications of this region abolish the inhibitory activity (6, 10, 168, see Ref. 174),and supports the suggestion (6, 10, 174)that the regulatory subunit inhibits the catalytic subunit by acting as a “substrate analog” of high affinity which shields the protein or peptide substrate binding site of the catalytic subunit. The autophosphorylation site could be the substrate analog site in question. From the magnetic resonance studies of Granot and co-workers ( 9 ) ,the type I1 regulatory subunit might be considered to be a “dead-end” substrate since the respective holoenzyme binds MgADP with high affinity. Type I isozyme does not undergo autophosphorylation, but it does contain a homologous autophosphorylation analog site (80) that could serve a similar purpose. In addition to the autophosphor-


65

m

@ I

U

I

FIG.4. Formation of the functionally competent dimer by trypsin treatment of the tetrameric holoenzyrne.

ylation primary sequence of type 11, the inhibitory domain must also include a structure(s) of higher order since heat denaturation destroys the regulatory subunit inhibitory activity, but not its ability to act as a substrate for autophosphorylation (174). With regard to this second postulated structure, since autophosphorylation can take place, the regulatory subunit does not block the MgATP-binding site of the catalytic subunit, but it could block the transfer of phosphate from ATP to substrate (10). It follows that the inhibitory action of the regulatory subunit could be exerted in two ways. First, it may act as a competitive inhibitor of high affinity; and second, it could block the ability of catalytic subunit to transfer phosphate from ATP to substrate. Several isoforms of a heat-stable protein kinase inhibitor of high specificity for the CAMP-dependent protein kinase are present in mammalian tissues (169171). Although the physiological role of this inhibitor has not been established conclusively, it has been quite useful in studies of mechanism and cellular role of the kinase. As is probably the case for the regulatory subunit, this inhibitor may act at least in part by serving as a competitive substrate analog of the CAMPdependent protein kinase (172). Scott ef al. have isolated a proteolytically derived fragment of the inhibitor that retains inhibitory activity (173). This fragment contains a substrate analog sequence as indicated by the presence of tandem arginines followed by alanine in the usual position of a phosphorylatable serine or threonine of natural protein kinase substrates. These investigators have synthesized a potent (K; = 0.8 @ peptide I)kinase inhibitor of an amino acid sequence corresponding to the first twenty amino acids of the proteolytically derived fragment. The amino acid sequence of this peptide resembles the sequence in the regulatory subunit which, as previously noted, is part of the domain that inhibits the catalytic subunit. Several findings are indicative that a ternary complex is formed between CAMP and holoenzyme during activation of the kinase (175-177). Although difficulties have been encountered in isolating and studying a stable ternary complex, several reports suggest that this is possible. Cobb et af. (64, 64a) separated and characterized two distinct peaks of homogeneous Type I1 bovine heart tetrameric holoenzymes. Using HPLC-DEAE chromatography, both peaks had similar molecular weights as calculated from their Stoke’s radii and szo.w

66


values or determined by denaturing electrophoresis. They contained equimolar amounts of regulatory and catalytic subunits and were highly CAMP-dependent. One of these peaks was identified as a relatively stable ternary complex which appeared to be holoenzyme (R2C2)with approximately half-filled cAMP binding sites. This ternary complex was further distinguished from the CAMP-deficient holoenzyme by an enhanced positive cooperativity in nonequilibrium [3H]cAMP binding and in kinase activation. This complex could represent an enzyme form that is primed for activation. At least two other reports have provided evidence for the presence of a ternary complex between cAMP and a trimeric form of CAMP-dependent protein kinase. Rangel-Aldao and Rosen (122a) and Connelly ef al. (83) identified an intermediate assigned an R,C structure. In contrast to the ternary complex reported by Cobb er al. (64, 64a), these trimer complexes were formed during reassociation of catalytic and regulatory subunits. The significance of these complexes remain to be fully determined. It is assumed that CAMPbinds in a step-wise manner to the four binding sites of the holoenzyme. The binding exhibits positive cooperativity with reported Hill constants of 1.6-1.8 for kinase activation (10). More direct demonstration of positive cooperativity is by the use of site-selective cAMP analogs. These investigations have shown that binding of cyclic nucleotide to either of the two intrasubunit sites (sites 1 and 2) has a strong stimulatory effect on binding to the other site (109). There may be several mechanisms for these stimulations. Although it cannot be ruled out that all four binding sites of the holoenzyme are accessible to the first mole of CAMP, the results of several experiments using type I1 suggest that binding to site 2 is blocked by the presence of catalytic subunit, but this site is made available by site 1 occupancy (12, 109). Thus, cAMP probably binds first to site 1, which then stimulates site 2 to bind. Using the free regulatory subunit of type I, it has been found that when cAMP occupies site 2, the dissociation of cAMP from site 1 is retarded (179). These characteristics could explain at least in part the observed stimulation of binding to both sites 1 and 2 noted above. The additional finding, using the free regulatory subunit of type I, that binding to site 2 of one subunit retards dissociation of cAMP from site 2 of the other subunit could also explain part of the stimulation (15). This latter intersubunit stimulatory effect is probably minor using type I1 since the cooperative binding is essentially unaltered in the dimeric (RC) holoenzyme as compared with the native tetramer (R2C2)( 2 ) . However, it has been calculated that an intersubunit positive cooperativity is necessary in order to explain the high Hill constant for cAMP activation of the protein kinase (180). The precise mechanism by which binding of cAMP to the enzyme causes activation is not yet clear. At the time of the discovery of the two intrasubunit CAMP-binding sites, it was suggested that, because of the high affinity of each, both are involved in activating the kinase (6). Subsequent experimental evidence has verified this claim (111, 181). The exact involvement of site 1 or site 2, and

3. CYCLIC NUCLEOTIDE-DEPENDENTPROTEIN KINASES

67

of the binding of the four total cAMP molecules to this enzyme in the activation process is not certain, however. Binding of cAMP to one of the sites could simply serve the function of stimulating the other site to bind without being directly involved in kinase activation. From the results of fluorescence-polarization spectroscopy studies, Seville er al. have concluded that the binding of two cAMP molecules to the holoenzymes is sufficient to cause the dissociation of both catalytic subunits (182). Studies of cGMP-dependentprotein kinase suggest that at least binding to site 1 of this enzyme is directly involved in partially activating the kinase (21). Binding to site 2 causes an additional increase in enzyme activity. Even though the CAMP-dependent protein kinase is a homologue of this enzyme, it cannot be assumed that it behaves identically with respect to kinase activation. Since the functional unit of the enzyme is a hetero-dimer (RC), this implies that one C at a time is replaced from the native tetramer (R2C2)during activation by CAMP. This suggests the existence of an intermediate form (R2C) (83, 122a). The process of CAMP-dependent protein kinase inactivation has also been investigated. It has been known for many years that the catalytic subunit induces release of cAMP from the R-CAMP complex (183). For type I enzyme, this process is stimulated by the presence of MgATP (183). 0greid and Doskeland have examined the effect of the catalytic subunit on the release of cAMP from each of the binding sites (184). From the experimental results, it has been determined that cAMP is first released from site 2, and that site 2 vacancy retards the release of cAMP from site 1. It has been also determined that both of the cAMP molecules are released from one subunit of the dimer before cAMP is released from the other subunit.

B. MECHANISM OF CATALYTIC SUBUNIT ACTION The catalytic subunit action has been reviewed extensively (23) and will be addressed only briefly in this chapter. The available evidence indicates that the catalytic subunit has a single active site on each monomer. The monomer incorporates a single substituent when photoaffinity labeled either with the 2 ' 3 ' dialdehyde derivative of ATP (185) or with p-fluorosulfonylbenzoyl adenosine (186, 187). In the latter case, the label is attached to lysine-71 of the primary amino acid sequence of the catalytic subunit. The sequence of this region of the molecule, which presumably represents at least part of the active site, is Leu-Val-Lys-His-Lys-Glu-Thr-Gly-Asn-His-Phe-Ala-Met-Lys* -1le-Leu-AspLys-Glu-Lys-Val-Val-Lys-Leu-Lys-Gln-Ile. The catalytic subunit can also be labeled in a single site per monomer with o-phthalaldehyde (188)or with peptide substrates containing reactive groups such as 3-nitro-2-pyridinesulfenyl (189). It is well known that sulfhydryl modifying agents abolish activity of the catalytic subunit and that MgATP protects against these agents (54, 55, 77, 79,

68


190). Of two -SH groups in the catalytic subunit, cysteine-199 appears to be located in the active site (189). The amino acid sequence around this residue is GIy-Arg-Thr-Thr-Thr-Leu-Cys* -G1y-Thr-Pro-Glu-Tyr-Leu-Ala (79). From the observation that catalytic subunit modified at both lysine-71 and cysteine- 199 does not form an isoindole derivative on treatment with o-phthalaldehyde, together with consideration of transition energy measurements on the o-phthaladehyde modified enzyme, the lysine and cysteine residues in the active site are about 3 A apart and are proposed to be in a hydrophobic environment (188). From NMR (191) and ATP analog specificity studies (192), it is known that the ATP is bound in the anri-conformation in the active site. Data from several laboratories (191) indicate that the catalytic subunit utilizes P,y-bidentate MgATP with the geometry shown below. The coordinates of the binding of this complex in the active site have been mapped (191).

There have been numerous inconsistencies concerning the kinetic mechanism by which the catalytic subunit catalyzes the transfer of the y-phosphate of ATP to substrate (193-198). Whitehouse et al. have carefully analyzed the reaction mechanism using the enzyme, MgATP, peptide substrate, substrate analogs, and protein kinase inhibitor (198). They have concluded that the steady-state kinetics follow an ordered bi bi mechanism in which ATP binds first. The terminal anhydride bond is then proposed to undergo a conformational change, induced either as a consequence of ATP binding per se or by the subsequent binding of protein substrate. Catalysis then ensues, leading to the sequential release of phospho-substrate and ADP. The substrate specificities of CAMP- and cGMP-dependent protein kinases have been reviewed extensively (13, 14, 31, 38, 39). Although there are slight differences in virro in specificity, in general, they both require a pair of basic amino acids on the amino terminal side of phosphorylated serines or threonines. Small peptides with amino acid sequences of this type are excellent substrates for the enzymes. Granot et al. determined by a process of elimination that if a particular conformation of protein or peptide substrate is required in the active site, then it is probably a coil (199). From results of induced circular dichroism, Reed and Kinzel have concluded that binding of the protein substrate results in a . change apparently conformational change at the ATP-binding site ( 2 0 0 ~ )This takes place in at least two steps, one dependent on the presence of a phosphorylatable serine or threonine in the substrate, and the other dependent on the pair of basic amino acids. The authors have also postulated from the results of competition experiments that the binding site closes over the substrate protein


69

following the initial binding. The catalytic subunit from porcine heart has been crystallized, which should allow for more detailed studies of its structure and mechanism (200).

V. Biological Role of Protein Kinases A. CRITERIA TO ESTABLISH BIOLOGICAL ROLE Several approaches have been used to elucidate the roles of cAMP and the CAMP-dependent protein kinase in cell function. Sutherland ( 4 ) first proposed criteria that should be satisfied to justify that a given hormone acts through CAMP. These criteria include the demonstration that the hormone activates adenylate cyclase and elevates the intraceliular level of CAMP. Additionally, the response should be potentiated by phosphodiesterase inhibitors and should be mimicked by the addition of exogenous cAMP or cAMP analogs. The discovery of the CAMP-dependent protein kinase (5,56) led to the hypothesis that all of the effects of cAMP are mediated by phosphotransferase reactions catalyzed by this enzyme. Subsequently, Krebs proposed a set of criteria that should be satisfied before a CAMP-mediated response could be established to be carried out by the CAMP-dependent protein kinase (201). These criteria have also been reevaluated (11, 13, 37). It should be demonstrated that a protein substrate, which is shown to be involved in the response, can be stoichiometrically phosphorylated (and dephosphorylated) at an appropriate rate in vitro and in vivo and that this is correlated with an appropriate coordinant change in the function of the substrate. It is the in vivo aspects of these criteria that are the most difficult to satisfy. In many instances the measurement of the CAMP-dependent protein kinase activity ratio (kinase activity in the absence of cAMP divided by activity in the presence of CAMP) in crude extracts from hormone-stimulated intact cells or tissues is a valid indication of the activation state of the enzyme. The cooperative binding of cAMP to the regulatory subunit causes a cooperative activation of protein kinase (12, 85, 112). It can therefore be expected that small changes in the level of cAMP will result in large changes in the activation of protein kinase. Consequently, since this enzyme is at a pivotal point in the overall metabolic regulation of the cell, an accurate determination of the protein kinase activity ratio is probably a more informative indicator of the potential of a tissue to respond to hormone stimulation than is determination of the level of CAMP. The difficulty in detecting small changes in the level of cAMP may be partially responsible for some controversies concerning the involvement of cAMP in mediating certain physiological responses. Although activity ratio measurements can be a useful and valid method for evaluating hormone action and protein kinase activation, it is critical that these studies are carried out carefully and that certain pitfalls are recognized and avoided. Since Flockhart and Corbin have

70


reviewed the method and potential pitfalls (10, 202), only brief comments and later developments are included here. The most fundamental requirement for valid activity ratio determination concerns maintaining the intracellular protein kinase activation state following homogenization. Postextraction activation or inactivation causes an overestimation or underestimation of the activity ratio, respectively. Both of these potential artifacts can be controlled (10, 202). The most important advance in this procedure is the use of a synthetic peptide as substrate (203).In general, use of this substrate instead of histone is recommended since it is more specific and sensitive, does not cause protein kinase dissociation, is not inhibited by NaCl under any recommended homogenizing technique (S. J. Beebe and J. D. Corbin, unpublished), and is a poor substrate for phosphoprotein phosphatases. The cGMP-dependent protein kinase also phosphorylates this substrate but this generally is not a problem since this kinase activity is a minor component of most tissues. Another approach used to define the biological role of the CAMP-dependent protein kinase is genetic analysis using mutant cells that have a single lesion in the pathway of cAMP action, preferably an altered or absent protein kinase [see Ref. (204) for a review]. If the CAMP-dependent protein kinase were required for a given response, a mutant deficient in protein kinase would be unable to elicit the response. Such mutant cells provide advantages since, in contrast to the wildtype cells, their growth is not inhibited by cAMP or cAMP analogs. Several criteria have been used to determine that a phenotypic variation is due to a mutation (205).Cyclic AMP-resistant cells must arise spontaneously, be clonally inherited, and the phenotype must be stable. In addition, the mutation must arise at a frequency consistent with mutation in microbial systems and increase in frequency in the presence of known bacterial mutagens. Conclusions derived from these studies are valid only if the mutation in question is not a pleiotropic one affecting multiple cellular functions. Several other approaches have been used to establish that the protein kinase mediates a given response. The direct introduction of the catalytic subunit of protein kinase or its specific heat-stable inhibitor into a cell can be achieved by microinjection. This approach was used in Xenopus oocytes (206). Although the technique is generally restricted to relatively large cells, cAMP and the subunits of the CAMP-dependent protein kinase were successfully injected into isolated guinea pig ventricular myocytes (207, 208). A similar method was used to fuse vesicles with isolated or cultured cells. Culpepper and Liu (209) fused CAMPcontaining and 8-azido-CAMP-containingvesicles with H4 and H35 hepatoma cells, and Boney et af. (210) incorporated catalytic subunit and protein kinase inhibitor into H35 hepatoma cells using protein-loaded human erythrocyte ghosts. Bkaily and Sperelakis (211)fused cultured myocytes with phosphotidylcholine liposomes containing catalytic subunit and protein kinase inhibitor.


71

Another approach that has been used is incubation of intact cells or tissues in the presence and absence of hormone or other agent. The potential protein substrate is then isolated and phosphorylated in v i m by addition of [32P]ATP and the catalytic subunit of protein kinase. The decrease in 32P content of the protein substrate that occurred in the presence of the agent compared to the control can be assumed to have occurred in vivo (212). A technique has been used that is based on a variation of the Sutherland criteria requiring that cAMP analogs mimic a given CAMP-mediated response. It is basically a shortcut to the Krebs criteria, which were designed to establish that a given response is mediated by the CAMP-dependent protein kinase. The procedure is modified to account for the presence of two different intrasubunit cyclic nucleotide-binding sites on the regulatory subunit of protein kinase. Intact cells are incubated with two cAMP analogs either alone or in combination (113, 114, 1141). The procedure takes advantage of two properties of cAMP analogs, (a) the passage of analogs across cell membranes and the direct activation of the protein kinase; and ( b ) the site selectivity of the two different analogs each of which binds to one or the other of two distinct intrachain binding sites (Fig. 1). This occurs in such a way as to facilitate the positively cooperative activation of protein kinase. The advantages are that it is highly specific to the protein kinase isozymes and it does not require the measurement of cAMP binding to or activation of protein kinase. Instead, measurements of the response are made due to protein kinase activation in the intact tissue. Furthermore, proper use of this procedure can potentially differentiate between type I and type I1 protein kinasemediated metabolic events. Since the technique measures the metabolic response itself, it does not specifically require the identification of the phosphorylated protein which leads to the response. Although it does establish that the response is due to protein kinase activation, it does not specifically correlate the response to a phosphorylation reaction. For example, it does not rule out that the free regulatory subunit is involved in the response. In addition, the use of analogs cannot necessarily predict hormone responses, but it can modify them. For instance, cAMP analogs have been shown to block the endogenous cAMP elevation in glucagon-stimulated hepatocytes (213). Furthermore, the insulin blockade of adipocyte lipolysis or hepatocyte glycogenolysis depends on the cAMP analog used to elicit the response (214). Several other procedures have been used to correlate a metabolic response with the activation of one or the other of the protein kinase isozymes. These are discussed in Section V,C.

B. DISTRIBUTION OF ISOZYMES An examination of the relative tissue and species distribution of the isozymes of the cAMP kinase does not provide proof for specific roles for the isozymes but does present some interesting considerations. While the ratios of the regulatory

72


subunit to the catalytic subunit of the CAMP-dependent protein kinase are relatively constant among various tissues (90, 216), some tissues contain either predominantly type I or type I1 protein kinase while others contain an equal mixture of the two isozymes (7, 14). For instance the brain, stomach mucosa, and adipose tissue from many species contain predominantly the type I1 isozyme while rabbit psoas muscle, bovine corpus lutea, rat testes, and bovine neutrophils contain primarily the type I protein kinase. Rabbit soleus muscle, rat liver, rabbit reticulocytes and erythrocytes, and human neutrophils contain a mixture of both isozymes. Cardiac tissues are particularly interesting regarding isozymes. While bovine and guinea pig heart contain predominantly type 11, rat and mouse heart contain type I and rabbit and human heart contain an equal mixture of the isozymes (215). In addition, cardiac tissue has been shown to contain a CAMPdependent protein kinase in the particulate fraction (216), which has been reported to be primarily associated with sarcolemma (217-223) and sarcoplasmic reticulum (224-228). Although in most cells and tissues the enzyme is predominantly in the soluble fraction, membrane-bound CAMP-dependent kinase activity has also been found in brain (229), erythrocytes (230), corpus luteum (231), sperm (232, 233), and thyroid (116).The membrane-associated kinase is usually found to be the type I1 isozyme, but the type I isozyme has been reported in membranes from erythrocytes (230), sperm (232, 233), and thyroid (116). The holoenzyme is apparently attached to particulate material by its regulatory subunit (216). However, the catalytic subunit has been reported to nonspecifically bind to membranes under conditions of low ionic strength (234).Both the particulate-bound holoenzyme and the regulatory subunit are readily solubilized by a number of conditions, which, according to the criteria of Singer (235),classifies them as “peripheral,” as opposed to “integral” proteins. Immunochemical studies, using antibodies that are specific for the regulatory subunit of the cyclic nucleotide-dependentprotein kinases, have allowed a more precise localization of the isozymes and subunits. The advantages and limitations of these techniques have been reviewed (236-238). An increase in nuclear protein kinase has been reported in regenerating liver (239), ACTH-stimulated adrenal medulla (240),and growing human breast cancer cells (241). Van Sande et af. (242)reported that the type I1 regulatory subunit is present in the nucleus of thyroid follicular cells and the type I regulatory subunit, catalytic subunit, and cGMP-dependent protein kinase are primarily in the cytoplasm and associated with the apical membrane. Jungmann and co-workers, using an indirect colloidal immunogold technique (243), found that only the catalytic subunit is present in the nucleus of glucagon- or dibutyryl-CAMP-stimulated hepatocytes (244), but the catalytic subunit and both type I and type I1 regulatory subunits are in the nuclei of regenerating liver cells (245). Fletcher and Byus (246, 247), using a fluorescein-conjugated heat-stable protein kinase inhibitor, determined that catalytic subunit appears rapidly (5-15 min) in the cytoplasm and nucleolus but


73

slowly (1 h) in the nucleus of glucagon- or dibutyryl-CAMP-stimulated H35 hepatoma cells. In contrast, Murtaugh et al. (248) were unable to detect any apparent redistribution of catalytic subunit in 8-bromo-CAMP-stimulated CHO cells and observed a diffuse staining pattern in several other cultured cell types. It is presently not clear if these differences are due to variations in basic mechanisms among cell types or to technical differences. The type I1 isozyme has been reported to be bound to a number of proteins, including microtubule-associated protein 2, a brain cytoskeletal protein (249-252); the mitotic spindle and nuclei of human breast cancer cells during different phases of growth (253);calcineurin (254), a calcium-calmodulin-activatedprotein phosphatase (255); P75, an unidentified brain protein with M, = 75,000 (256);and to some other brain proteins which are distinct from several type I1 regulatory subunit-bound proteins in heart (252). It is interesting that many of these examples are calcium-calmodulinbinding proteins which serve as substrates for the CAMP-dependent protein kinase. The cGMP-dependent protein kinase was first discovered in arthropod tissues (257) and was subsequently demonstrated in mammalian tissues (258). In contrast to the CAMPkinase, the cGMP-dependent protein kinase represents a minor component in most cells and is more restricted in its distribution (259-261). However, the cGMP kinase is found in significant concentrations in heart, lung, intestine, adrenal cortex, cerebellum, and smooth muscle tissue (25). Immunocytochemical studies localize the cGMP kinase in the cerebellum and the smooth muscle cells of major and minor blood vessels, intestinal wall and respiratory tract (262).It is presently not clear whether there are specialized functions in various tissues which require a specific cyclic nucleotide-dependent protein kinase isozyme or if there is a special advantage for a given tissue to contain a specific isozyme or isozyme mixture. An alternative explanation, that either isozyme can serve the physiological function equally well, and that the presence of different isozyme ratios in the various tissues is only fortuitous, has not been proved. However, the cGMP and the cGMP-dependent protein kinase has been reported to mediate the effects of certain agents, such as nitroglycerin and atrial natriuretic factor, on smooth muscle relaxation and kidney functions (262~2, 262b). ACTIVATION OF CYCLIC AMP-DEPENDENT C. SELECTIVE PROTEIN KINASEISOZYMES

Several factors regarding the physiological roles of the isozymes can be considered. One of these relates to the subunit structure of the kinases and the proposed mechanism of regulation of metabolism in the cell. As already described, both isozymes are composed of an inhibitory regulatory subunit dimer and two catalytic subunits. It is the difference between the regulatory subunits

74


that determine the isozyme type. According to the prevailing view, in higher organisms all of the effects of cAMP are mediated by the CAMP-dependent protein kinase and all regulatory processes are believed to be brought about by phosphorylation reactions catalyzed by the catalytic subunit of the enzyme. Since the catalytic subunit is identical in both isozymes, it can be argued that either isozyme can mediate the responses initiated by elevations in intracellular concentrations of CAMP. This does not exclude the possibility that an isozyme may be compartmentalized so that a single isozyme is activated by hormone stimulation. One should interpret with caution the finding that the type I isozyme is activated in vitro at lower concentrations of cAMP than is the type I1 isozyme. Activation of type I isozyme is inhibited by physiological concentrations of MgATP and autophosphorylation of the type I1 regulatory subunit inhibits subunit reassociation. Hofmann presented in vitro evidence indicating that in the presence of MgATP and sodium chloride the concentration of cAMP required to dissociate each isozyme is similar (57).Other factors such as dissociation of the isozymes by basic protein substrates (263, 264) and differential salt activation of the isozymes may be important in vivo. The extent of activation will also depend upon the relative concentrations of the regulatory and catalytic subunit present (265). As previously mentioned enzyme compartmentalization may be an important factor if selective activation of a single isozyme occurs. Buxton and Brunton (266)demonstrated selective activation of protein kinase based on the subcellular localization of the enzyme in a homogeneous population of cardiac myocytes. Specifically, isoproterenol and prostaglandin E, cause an elevation of cytosolic cAMP and an activation of cytosolic protein kinase. However, isoproterenol, but not prostaglandin E l , causes an elevation of particulate CAMP, a translocation of protein kinase activity from the particulate to the cytosolic fraction, and an activation of glycogen phosphorylase. This isoproterenol-specific response is rapid and is temporally related to the phosphorylase activation. That differential calcium availability is responsible for the observed differences is largely ruled out (266).These results suggest that the P-receptor adenylate cyclase, protein kinase, and phosphorylase are spatially isolated from the prostaglandin El system in cardiac myocytes. This agonist-specific protein kinase activation has been demonstrated in isolated, perfused hearts from several species (267-269). In the following discussion, reports supporting selective isozyme activation are examined and then reports of simultaneous activation of the isozymes are reviewed. Selective activation of the CAMP-dependent protein kinase isozymes has been reported in intact organisms and in intact cells. Schwoch (270)evaluated the extent of activation of the type I and type I1 protein kinases in rat liver after the animals were injected with glucagon. This analysis was based on DEAE-cellulose separation of inactive holoenzyme and free catalytic subunit and the property of the type I1 isozyme to rapidly associate in low ionic strength.


75

Although the technique has a potential for artifactual enzyme activation (see Section II,A), it was one of the first studies designed to evaluate the physiological significance of two CAMP-dependent protein kinase isozymes. The data suggested a sequential activation of type I and then type I1 protein kinase. More specifically, 2 min after glucagon injection the type I isozyme was completely activated for at least 60 min while the type I1 isozyme was only partially activated. The type I1 kinase remained active for only about 30 min while the levels of CAMP were relatively high. Byus et al. (271) also found that a sequential activation of type I and type I1 protein kinase occurs when normal hepatocytes are incubated with glucagon or dibutyryl-CAMP. They used a method to distinguish type I and type I1 kinase activation based on separation of free catalytic subunit, type I and type I1 protein kinase using C6-aminoalkyl agarose chromatography. Maximal glycogenolysis was correlated with a selective activation of type I protein kinase, which was predominantly activated by the lowest effective concentrations of either agonist. Hunzicker-Dunn (272) also used a technique that separated subunits of CAMPdependent protein kinase on DEAE-cellulose. Data from this study suggested a preferential activation of the type I isozyme in the corpora lutea obtained from ovaries of 4-day pseudopregnant rabbits treated with a single injection of human chorionic gonadotropin. Livesey and co-workers (273) developed a rapid batch elution method for separating the two isozymes of protein kinase on DEAE-cellulose columns. In these studies rigorous attempts were made to exclude the possibility of postextraction activation of protein kinase. One study evaluated the effects of parathyroid hormones and prostaglandin E, on the activation of type I and type I1 protein kinase from normal and neoplastic osteocytes from rat calveria (273). These data suggested that the isozyme response not only is specific for a particular hormone effector but also depends upon the cell type. Parathyroid hormone predominantly activated the type I isozyme in the neoplastic osteocytes but activates both isozymes to the same extent in the normal cells. Prostaglandin E, also caused a predominantly type I isozyme activation in the malignant cells but specifically activated the type I1 isozyme in the normal calveria cells. This technique was modified and validated to evaluate the effects of conditions and prostaglandin E, on protein kinase activation in two human breast cancer cells lines, T47D and MCF7 (274, 275). In both of these cells, calcitonin selectively activated the type I1 isozyme, however, the duration of the response was different. While type I1 activation in the MCF7 cells was transient over 4-6 h, it was persistent in the T47D cells for at least 24 h. Mizuno et al. (276) analyzed the cytosolic protein kinases from the submandibular glands of control and isoproterenol-treated rats. Using DEAE-cellulose, they concluded that 10 min after isoproterenol injection, the type I1 isozyme readily dissociated into regulatory and catalytic subunits which are separated on

76


the column. The type I isozyme eluted from the column at 0.1 M KCI but had an activity ratio of about 1.0; they concluded that the type I isozyme was less dissociable than the type I1 isozyme, but was highly active in an undissociated form. Therefore, activation of both isozymes correlated with salivary secretion but the type I remains active for as long as an hour while the activity ratio of the type I1 isozyme decreases slightly. This latter result is similar to the timedependent behavior of glucagon-stimulated rat liver protein kinase isozymes reported by Schwoch (270). Chew (277) investigated histamine-stimulated acid secretion in parietal cells isolated from rabbit gastric mucosa. These cells contained the type I isozyme in the cytosol and the type I1 kinase in the cytosol and particulate-fraction. Histamine activated only the type I isozyme while forskolin activated both isozymes. This conclusion was obtained from results of DEAE-cellulose chromatography, differential isozyme reassociation after cAMP removal by Sephadex G-25 chromatography, and 8-a~ido-[~~P]cAMP photoaffinity labeling of crude extracts. The results suggest that the type I isozyme mediates histaminestimulated acid secretion and that this isozyme is compartmentalized with the histamine H, receptor-coupled adenylate cyclase. Other techniques have been used to evaluate the biological role and the physiological significance of the two isozymes of the CAMP-dependent protein kinase. Litvin and co-workers (278) used an approach based on specific antibody precipitation of type I or type I1 protein kinase. They demonstrated a parallel release of ACTH and a selective type I protein kinase activation when AtT20 mouse pituitary tumor cells were stimulated with corticotropin releasing factor. Maximal ACTH release was seen with a 2-fold increase in cAMP and only a slight activation of the type I1 isozyme. When 0.5 mM 3-methylisobutylxanthine was added, no further increase in ACTH release occurred and the type 11 isozyme is now activated by 50%. 3-Methylisobutylxanthine alone (0.5 mM) fully activates type I, while type I1 was activated by about 25%. The earliest report of simultaneous activation of types I and I1 in a tissue was that of Corbin and Keely (215). This study was done using epinephrine-perfused hearts from several mammalian species which were known to possess different relative amounts of the two isozymes. From examination of the degree of reversal of the hormone effect by Sephadex G-25 chromatography, no apparent selectivity was observed and it was concluded that both isozymes are activated by epinephrine. Ekanger et al. (279) developed an approach in which endogeneous cAMP bound to the regulatory subunit of type I or type I1 protein kinase was quantitated after specific adsorption to protein A-agarose coated with antibodies directed against the respective isozymes. While the work of Schwoch with rat liver (270) and Byus et al. with isolated hepatocytes (271) indicated a preferential activation of type I protein kinase, Ekanger and co-workers demonstrated that both type I


77

and type I1 isozymes were equally activated in isolated hepatocytes as the concentration of glucagon is varied over a wide range. As discussed in Section II,A, certain cAMP analogs are selective for one or the other of two intrasubunit cyclic nucleotide binding sites and cAMP analog combinations can distinguish between synergistic activation of type I and type I1 protein kinase (Fig. 1). These in vitro findings have been extended to demonstrate a synergism of CAMP-mediated responses in a number of isolated cell and tissue preparations (113, 114, 1 1 4 ~ )It. has been demonstrated that the analogs used in these studies have no apparent special properties other than site selectivity (105, 107, 109, 111, 113). Furthermore, when hepatocyte glycogenolysis is stimulated by a series of equipotent concentrations of glucagon or 8-thioparachlorophenyl-CAMP [and the analog effectively removed (Ref. 2131, the degree of phosphorylase activation was closely correlated with the CAMP-dependent protein kinase activity ratio for both agonists (T. W. Gettys and J . D. Corbin, unpublished). Therefore, data demonstrating that a combination of site-selective cAMP analogs generate a synergism of a physiological response can be used to prove that protein kinase mediates the response. This technique also suggests that the cooperativity of cAMP binding and protein kinase activation measured in vitro are important to the function of protein kinase in vivo and that the cooperativity may be a mechanism of sensitivity amplification at the protein kinase step in the intact tissue (113). When cyclic nucleotide analogs are added to isolated cells, they cross the cell membrane, bypass the hormone receptor-adenylate cyclase system and directly activate the CAMP-dependent protein kinase isozymes (113, 114, 114a, 214). When a site-1- and a site-2-selective analog are added in combination to intact cells synergism of a physiological response occurs only if the response is mediated by the CAMP-dependent protein kinase. Furthermore, if the type I protein kinase mediates the response, synergism should occur only when two analogs selective for site 1 and site 2, respectively, for this isozyme are combined (type I directed-analog pair), and if the type I1 isozyme mediates the response synergism should only occur when two analogs selective for site 1 and 2 of this isozyme are combined (type I1 directed-analog pair) (see Section II,A, and Fig. 1). It is important to use the analogs at relatively low concentrations, which allows for a wider window for examining synergistic effects (113, 114, 1 1 4 ~ )It. is also important to use the analogs at concentrations that result in a linear doseresponse relationship for the measured effect. Additionally, controls should be included to determine that a combination of two analogs that are selective for the same site do not cause a synergism of the response. This method, which is fully described elsewhere, has now been tested in a number of different systems (114a) including adipocyte lipolysis ( I 13), hepatocyte phosphorylase activation (114), increases in mRNA for phosphoenolpyruvate carboxykinase in continuous cultures of H4IIE hepatoma cells (280),induc-

78


tion of LH receptors and increased progesterone synthesis in primary cultures of porcine granulosa cells (281), and thyroid hormone release in dog thyroid slices (J. Van Sande and J . E. Dumont, personal communication). This technique was first tested by measuring adipocyte lipolysis (113). Adipocytes contain predominately, if not exclusively, the type I1 isozyme (7). Synergism of protein kinase activation in v i m and glycerol release from intact cells was observed only with type I1 directed-analog pair. Synergism did not occur for either process with type I directed-analog pair, or with a combination of two site-I or two site-2 selective analogs. These data demonstrate that the type I1 isozyme mediates the lipolytic response. Since the first demonstration of synergism in intact cells using pairs of siteselective cAMP analogs, new analogs have been developed that, when combined with another analog selective for the opposite site, provide a more selective synergistic activation of only one isozyme. One of these analogs is 8-piperidinocAMP (104). In addition to using N6-benzoyl- and 8-aminohexylamino-CAMPas a type I directed-analog pair and N6-benzoyl- and 8-thiomethyl-CAMPas a type I1 directed-analog pair, further experiments were carried out using 8-piperidinoand 8-piperidino-CAMP as a type I directed-pair and N6-benzoyl- and 8piperidino-CAMP as a type I1 directed-pair (see Fig. 1). In these experiments, phosphorylase activation was measured in various cell types which contain different protein kinase isozyme ratios. Bovine neutrophils (85% type I), rat adipocytes (>95% type II), and rat hepatocytes (type I = type 11) were used. For all isozymes tested in v i m , type I directed-analog pairs synergistically activate only type I isozymes while type I1 directed-analog pairs synergistically activate only type I1 isozymes. In close agreement with these results, bovine neutrophil phosphorylase was synergistically activated almost exclusively by type I directed-analog pairs, adipocyte phosphorylase was synergistically activated almost exclusively by type I1 directed-analog pairs, and rat hepatocytes were synergistically activated by both type I directed- and type I1 directed-analog pairs (114). These data indicate that either type I or type I1 protein kinase can regulate phosphorylase activation. These techniques have been used to determine whether the CAMP-dependent protein kinase has a regulatory role at the nuclear level in H4IIE hepatoma cells. Since it is known that cAMP analogs increase the amount of mRNA for phosphoenolpyruvate carboxykinase (PEPCK) in these cells (283), it was of interest to see if cAMP analog combinations could act synergistically to produce this response. It was also possible to test if a synergism of PEPCK gene transcription could be demonstrated. To test these possibilities, various site-l- and site-2selective cAMP analogs were added alone (in the linear dose-response range) and in combination to H4II cell type I1 CAMP-dependent protein kinase in vitro and to H4 cells in culture. Determinations were then made of the extent of synergism for the isolated activation, increase in mRNAPEPCK,and PEPCK gene transcrip-


79

tion. All analog pairs that resulted in a synergistic activation of protein kinase also resulted in a synergistic increase in mRNAPEPCKand PEPCK gene transcription. Synergism, as quantitated by ratio of responses of an analog pair divided by sum of single analog responses, is approximately 2 for protein kinase activation, 3-4 for the increase in mRNAPEPCK,and 10-12 for PEPCK gene transcription. For all responses tested, synergism does not occur when two site- 1 -selective or two site-2-selective analogs are combined. The occurrence of synergism only when a site-1- and a site-2-selective analog are combined is a protein kinase-specific characteristic. These results clearly indicate the involvement of protein kinase in the cAMP regulation of PEPCK gene transcription. It is presently not clear if this effect is due to a phosphorylation event or to some other mechanism such as the CAMP-bound regulatory subunit acting in a manner analogous to the E. coli catabolite gene activator protein (CAP). Segaloff et af.(281) used these site-selective cAMP analog combinations in primary cultures of porcine granulosa cells to test whether or not induction of LH receptors and the increase in progesterone synthesis are CAMP-dependent protein kinase-mediated events. Photoaffinity 8-a~ido-[~*P]cAMP labeling of cell extracts suggested that both regulatory subunit isozymes are present. Two specifically labeled bands with apparent M, = 56,000 and 49,000 comigrate with homogeneous rabbit skeletal muscle type I and type I1 regulatory subunit standards. The type I1 isozyme is induced 5- to 10-fold by cholera toxin and represents the predominant isozyme while type I represents only a minor band. Incubation of cells in primary culture with type I directed- and type I1 directedanalog pairs resulted in a synergism of LH receptor induction, as determined by 1251 human choriogonadotropin binding, and in a synergism in progesterone synthesis, as measured by radioimmunoassay. The synergism of both of these responses was much greater with a type I1 directed- than a type I directed-analog pair. No synergism of either response was observed with two site 1-selective analogs or with two site-2 selective analogs. These data indicate that both the induction of LH receptors and the increase in progesterone are mediated by the CAMP-dependent protein kinase and that the predominant type I1 isozyme is primarily responsible for both of these responses. This method is presently being tested by J . Van Sande and J. E. Dumont (personal communication), using thyroid hormone secretion in dog thyroid slices. Secretion was measured by butanol extraction of I 3 l I from the medium as a percentage of total radioactivity present in the slices at the beginning of a 5 h incubation. Both type I and type 11 protein kinase were present in the dog thyroid (242). When tissue slices were incubated with type I directed- and type I1 directed-analog pairs, synergism of thyroid hormone secretion was present to approximately the same extent for both isozyme synergistic pairs. This result suggests that both type I and type I1 isozymes mediate thyroid hormone secretion.

80


Data from studies using type I directed- and type I1 directed-analog pairs suggest that, for the various parameters measured, either isozyme can mediate CAMP-dependent protein kinase generated responses. The extent of the contribution of each of the isozymes seems to correlate approximately with the amount of each of the isozymes present. These conclusions appear to be valid for the limited number of experimental models tested, and this technique appears to be generally applicable. It should be pointed out that through the use of cAMP analogs, the hormone-receptor-adenylate cyclase system is bypassed. It is indeed possible that the intact cell is compartmentalized to the extent that the hormone is coupled to only a single kinase isozyme. Furthermore, although for the type I and type I1 isozymes tested, the defined analog pairs appear to specifically cause a synergistic activation of only a single protein kinase isozyme, it cannot be ruled out that exceptions to these generalizations may exist. This is a concern given data describing microheterogeneous subforms of the type I1 isozyme from various tissues and species (see Section 11,A). Similar microheterogeneous subforms of the type I isozyme may also exist. However, for the responses of type I and type I1 protein kinase activation tested in v i m , the defined cAMP analog combinations appear to be isozyme specific. This is based on experiments of type I and type I1 isozymes from rabbit skeletal muscle, rat heart, rat hepatocyte, and bovine neutrophil, as well as the type I1 isozyme from rat liver plasma membrane, rat adipocytes, cultured rat hepatoma (H4IIE) cells, dog thyroid, and bovine heart. It should be pointed out that in spite of the microheterogeneous differences between the type I1 isozymes from bovine heart, rat hepatocyte, and rat adipocyte, similar effects of the cAMP analog combinations are observed with each isozymic form. D.

USE OF

CYCLIC NUCLEOTIDE ANALOGS IN INTACTCELLS

As mentioned previously, cyclic nucleotide derivatives were originally designed to be used in studies of the structural requirements of cyclic nucleotide activation of protein kinase, to develop analogs which are more potent than CAMP, and to develop antagonists of cAMP (281). These studies have been expanded to accommodate the presence of two types of CAMP-dependent protein as well as the cGMP-dependent protein kinases, designated type I and type I1 (3, kinase. Furthermore, it is now clear that the regulatory components of these three cyclic nucleotide-dependent protein kinase isozymes have two different intrasubunit cyclic nucleotide binding sites, termed site 1 or site B and site 2 or site A, which exchange bound, labeled cyclic nucleotide slowly and rapidly, respec, The development and uses of cyclic nucleotide tively ( 6 , 8 , 2 1 , 2 4 , 5 9 , 8 6105). analogs have closely paralleled and sometimes led the development of our understanding of the cyclic nucleotide-dependent protein kinases. In fact, derivatives of cyclic nucleotides which had been available for many years were used to


81

define and characterize the two different binding sites on the respective kinases (21, 105, 107, 108, 181). Consequently, studies are being directed toward understanding structural and functional differences among the cyclic nucleotidedependent protein kinase isozymes, finding analogs that are more potent activators of one or the other isozymes, and developing analogs that are selective for one or the other different intrasubunit binding sites. In addition to these in vifro studies, experiments applying the available knowledge concerning the mechanism of protein kinase isozyme activation, phosphodiesterase hydrolysis, and potential for analogs to be permeable to membranes have been and are being conducted using cyclic nucleotide analogs on a large number of in vivo preparations from whole animals to isolated cells. Because of the hundreds of analogs that have been synthesize, it has proved important to develop a systematic and rational approach to using analogs. Miller and Jastorff and their collaborators have synthesized and tested a large number of analogs (103, 285-292) that have proved useful for in v i m and in vivo studies. The work of these and others have greatly advanced our knowledge concerning the cyclic nucleotide-dependent protein kinases and their role in metabolic regulation. Both of the intrasubunit sites on the CAMP-dependent and cGMP-dependent protein kinases have high specificity for the ribose-phosphate moiety and lower specificity for the base moiety of the respective cyclic nucleotide (108). The CAMP-and the cGMP-dependent protein kinases distinguish between cAMP and cGMP by specific recognition of the respective bases of the molecules. The cAMP kinase isozymes are specific for a C6-amino group (290). Although the base is not required for activation, it is thought to interact with the regulatory subunit by hydrophobic and/or 7~ electron bonding interactions rather than by hydrogen bond interactions (290, 293, 294). Substitution at C6 results in reduced specificity for cAMP kinase activation and if the C6-amino group is changed to a C6-imino group, the resulting analog shows an increased specificity for the cGMP kinase (290, 291). As expected, the cGMP-dependent protein kinase appears to prefer both a C6-oxygen and a C2-amino group for activation. The C6-oxygen apparently accepts a proton from and the C2-amino donates a proton to the enzyme (290, 291). By systematically altering the cyclic nucleotide molecule, structural requirements for cyclic nucleotide binding and protein kinase activation have been partially defined. Evidence indicates that the primary determinant of cyclic nucleotide activation of the respective isozymes is the hydrophilic cyclic phosphate moiety (103, 285). Alterations in this portion of the molecule are generally not tolerated. Activation of both cAMP protein kinase isozymes requires a charged cyclic phosphate in addition to a 3' oxygen and a 2' hydroxyl group both in the rib0 conformation. These are probably important for hydrogen bonding (103, 289, 287). Additionally, the syn conformation of the cyclic nucleotide is pre-

82


ferred (287, 289) and the orientation of the purine-cyclic phosphate is critical (287, 289). Although the CAMP- and cGMP-dependent protein kinases are believed to be homologous proteins with highly conserved cyclic nucleotide binding sites, it is clear that differences do exist. These considerations also extend to the two isozymes of the CAMP-dependent protein kinase. Recent reports indicate that the type I1 isozymes from various tissues and species appear to be microheterogeneous (see Section II,A), and may contain differences in the binding sites. It should be kept in mind that the established structure-activity relationships for cyclic nucleotide recognition by the protein kinases are from a limited number of isozyme types. Cyclic AMP analogs have been used in a number of isolated cell and tissue preparations. The efficacy of cyclic nucleotide analogs as agonists of protein kinase-mediated responses in intact cells or in intact organisms is dependent upon the concentration of the analog at its site of action. The efficacy of all analogs tested for adipocyte lipolysis and hepatocyte phosphorylase activation are explained by consideration of at least three basic analog properties (113, 114a, 214) (favorable lipophilicity, low K , for protein kinase activation, and resistance to low K,,, phosphodiesterase hydrolysis). The lipid character and porosity of the cell membrane and other cellular constituents that could “trap the analog” should also be considered (214). Table I shows the concentration of analogs required for a half maximal activation of adipocyte lipolysis, hepatocyte phosphorylase a, and induction of mRNA for phosphoenolpyruvate carboxykinase in H4IIE hepatoma cells. There are only minor differences in the cyclic nucleotide specificities between the adipocyte and hepatocyte protein kinases and between the low K,,, phosphodiesterases from the respective cells types. Therefore, lipid content of cellular constituents, membrane porosity, and phosphodiesterase activity appear to be potentially important when comparing the efficacy of analogs in adipocytes and hepatocytes ( I 14a, 214). Adipocyte lipolysis is sensitive to analogs in the millimolar concentration range and hepatocyte phosphorylase is activated in the micromolar concentration range. Generally, hepatocytes are 100- 10,000 times more sensitive than adipocytes to cAMP analog stimulation when comparing these two physiological processes. The nuclear response measured in H4IIE hepatoma cells is intermediate in sensitivity to cAMP analogs. It is possible that other physiological responses in these cell types may show a different sensitivity to cAMP analogs. The sensitivity of a cellular response to cAMP analogs may also depend upon enzyme compartmentalization, the basal activation state of protein kinase (thereby affecting sensitivity amplification at the protein kinase step), and the potential for magnitude amplification (determined to some extent by the number of steps between the initial activation event and the final response). Free et al. (295) demonstrated that cAMP analogs were generally more potent in stimulating adrenal cell steroidogenesis than in stimulating adipocyte lipolysis. These studies

83

3. CYCLIC NUCLEOTIDE-DEPENDENTPROTEIN KINASES TABLE I CONCENTRATIONS OF CYCLIC NUCLEOTIDE ANALOGS PHOSPHOENOLPYRUVATE CARBOXYKINASE ACTIVATION

FOR

Cyclic nucleotide analog 8-Thioethyl 8-Aminomethyl N6-Diethyl 8-Bromo 8-Thio-p-chloropheny l 6-Thiomethy l 8-Thiomethyl 8-Amino 8-Thioisopropyl 8-Thiobenzyl 8-Thio-p-nitrobenzyl N6-Aminohex ylcarbamoylmethy l N6,02’-Dibutyryl N6-C~bamoylpropyl N6-Benzoyl N6-Butyvl 8-Hydroxy 8-Aminohexylamino 8-Aminobenzyl

ECSO Adipocyte lipolysis

EC50 Hepatocyte phos a

EGO mRNAPEPCK

(CLM)

(W)

(W)

0.3 6.0 I .3 0.1

>5,000 25

-

-

0.5 2.0 -

330 -

0.1

-

-

-

900 7,900 2,000 3,900 1,000 600 1,000 6,100 500 1,900 I ,200 10,000 1,100 500 700 1,900 900 8,600 >15,000

3.0 0.5 4.8 60 -

-

50

580 2,250 1,000 780

illustrate a degree of cellular specificity and suggest the possibility of using cAMP analogs in whole organisms to selectivity stimulate some cell types and not others. One of the originally defined objectives for the synthesis of cyclic nucleotide analogs was to find an antagonist of protein kinase (284). Some cAMP analogs have proven to be competitive antagonists of cAMP binding to the catabolite gene activator protein (CAP) and prevent specific DNA binding and stimulation of gene transcription in E. cofi (296, 297). For example, cGMP, cIMP, and N6monobutyryl-CAMP apparently interfere with optimal hydrogen bonding while the bulky groups of 8-bromo- and 8-thio-CAMPcause steric interferences. Presumably, these competitive antagonists and cAMP both bind to CAP but the antagonists do not induce the necessary conformational change required for specific DNA binding. The competitive antagonists of cAMP E . coli gene transcription mentioned above are agonists of protein kinase activation. The structural requirements for antagonism of protein kinase activation are therefore considerably different than for the CAP protein (298, 299).

84


Several reports using diastereomers of cyclic monophosphorothioate and cyclic monophosphodimethylamidate derivatives of cAMP [CAMPSand CAMPN(CH,),, respectively], which contain chiral phosphorus centers (RRp and SSp) suggest that some of these analogs may fulfill the criteria as useful protein kinase antagonists (300-306). These criteria include (a) binding to but not activating protein kinase; (b) having a relatively good affinity for protein kinase so that competition with cAMP at the cyclic nucleotide binding site is effective; and (c) having structural features that allow membrane penetration so that antagonists can be used in intact tissues. This latter criterion may vary for cells with different lipid characters. DeWit et al. (294, 305) used the diastereomeric forms of both of the above analogs to determine the characteristics of cyclic nucleotide binding to site 1 (stable site) on the type 1 isozyme from rabbit skeletal muscle. They observed that all the cAMP analogs tested, except Rp CAMPSand both enantiomers of CAMP-N(CH,),, have lower affinity for site 1 of the holoenzyrne than for the free regulatory subunit. This suggests that, unlike other analogs tested, these three analogs did not compete with cAMP for binding to the regulatory subunit. Furthermore, Rp CAMPS competed with cAMP for binding to the holoenzyme but was ineffective for protein kinase activation (294, 300, 305). O’Brian et al. (300) used Rp and Sp CAMPS to study the stereochemistry of cyclic nucleotide binding to and activation of the type I1 protein kinase. They reported that both enantiomers bind to and activate the kinase but the Sp conformer is much more potent for both of these actions than the Rp conformer. They also determined that the Sp conformer binds preferentially to site 2 while the Rp conformer binds preferentially to site 1. Botelho et al. have described a number of studies using Sp and Rp CAMPS in isolated hepatocytes (301, 302, 304, 306). Sp CAMPS is a full agonist that stimulates glucose production half-maximally and maximally at 0.8and 10 pM, respectively. Rp CAMPSappears to have no significant potency as an agonist of glucose production (301, 302), phosphorylase activation, or glycogen synthase inactivation (304). Instead, Rp CAMPS antagonizes these agonist-stimulated reactions. When 5 pM Rp CAMPSis present, 10-fold and 6-fold higher concentrations of Sp CAMPS and glucagon, respectively, are required for half-maximal glucose production. Similar increases in agonist concentrations are required for the activation of the CAMP-dependent protein kinase and phosphorylase, and inactivation of glycogen synthase. The putative antagonist (3 pM- 100 pM) produced a maximal inhibition of 50-74% when glucose production was stimulated by 1 nM glucagon, depending on the time of preincubation with Rp. Rp CAMPS appears to be an in vivo competitive protein kinase antagonist. It has been estimated that for Sp CAMPS-stimulated glucose production, a Rp CAMPS to Sp CAMPSmolar ratio of 3 and 30 are required for half-maximal and maximal inhibition, respectively. Further research using these protein kinase antagonists


85

and the development of others can provide useful and important tools to more completely define the role of protein kinase in the regulation of cell function. A potential complication of using cyclic nucleotide analogs in intact cells is a change in the intracellular cAMP concentration. For example, it is possible that analogs could serve as competitive substrates for phosphodiesterases and thereby increase the cAMP concentration. However, it has been shown that cAMP analog-stimulated hepatocytes have lower cAMP concentrations than unstimulated cells. The cAMP decrease appears to be mediated by a CAMP-dependent protein kinase mechanism (213). This suggests that through a feedback mechanism, such as phosphoidiesterase(s) stimulation and/or adenylate cyclase inhibition, the CAMP-dependent protein kinase can regulate its own activation state. The effects of insulin on cAMP analog-stimulated adipocyte lipolysis, hepatocyte glycogenolysis, and cardiac myocyte glycogenolysis and glycogenesis have been investigated (214, 307). In several mammalian tissues insulin is known to block most of the metabolic effects of hormones that elevate cAMP levels. If lowering of cAMP by phosphodiesterase activation is a necessary step in some of the effects of insulin, then this hormone may not work if the phosphodiesterase is prevented from catalyzing hydrolysis of cyclic nucleotides. This would be the case when cellular CAMP-dependent protein kinase is activated by phosphodiesterase-resistantcAMP analogs, instead of by either hormonal elevation of cAMP or by phosphodiesterase-sensitive cAMP analogs. Some, but not all, cAMP analogs are quite resistant to hydrolysis by phosphodiesterase (214). in cardiomyocytes insulin does not block the effects of the cAMP analogs tested on phosphorylase and glycogen synthase activity (307). In adipocytes and hepatocyges insulin blocks the metabolic effects of hormones that elevate CAMP, but when cAMP analogs are used, the results are different from those obtained in cardiomyocytes. Although insulin blocks the effects of some cAMP analogs it does not block the effects of analogs that are extremely resistant to hydrolysis by the low K,,,, hormone-sensitive phosphodiesterase (214). The results suggest that at least in adipocytes and hepatocytes insulin antagonism of the metabolic effects examined can be explained by phosphodiesterase activation.

E. THEROLE OF CYCLICAMP AND THE CYCLIC AMP-DEPENDENT PROTEIN KINASEISOZYMES I N THE CELL AND DIFFERENTIATION CYCLE,PROLIFERATION, The primary emphasis of this section is placed on cAMP and the CAMPdependent protein kinase since far more literature is available on this system as compared to cGMP and the cGMP-dependent protein kinase. Three levels of cyclic nucleotide involvement in the regulation of these developmental processes are discussed. The role of cAMP in these processes is only briefly reviewed here since this subject has been intensely studied over the years and has been the

86


target of several excellent reviews (308-318). The role of the CAMP-dependent protein kinase in these processes has been studied less frequently but has also been reviewed (239, 319, 320). The third level of involvement examines the possibility that specific cellular functions may be regulated by specific protein kinase isozymes. The literature in this field is inconclusive and often contradictory. It is difficult to arrive at definitive conclusions because methodological limitations require conservative interpretations (237, 312, 315, 318). Much of the early work was conducted without an appreciation for some of the pitfalls inherent in the experimental techniques. The events that regulate the cell cycle and determine if a cell will continue to proliferate or undergo differentiation are extremely complex. Research is being conducted against a background of incomplete and insufficient information. Several lines of evidence have accumulated to suggest a role for cAMP in the modulation of the cell cycle. One of the most fundamental observations supporting this is that the cAMP level in populations of synchronized cells oscillates during the cell cycle (311, 312, 314, 315, 317, 321-323). These cAMP fluctuations have a regular pattern and have been observed in a number of cell types, including normal and malignant cells, and using several different methods of cell synchronization. The most common finding is that the cAMP level is at a minimum during mitosis, gradually increases during G I , reaches a peak near the G , S border and declines again during the S phase. The G , phase is also often associated with a transient cAMP elevation. It is often seen that during the G I phase two peaks of cAMP are observed (311, 314, 318, 324-326). When two peaks occur, it appears to be the second peak that is required for DNA synthesis (318, 325, 327, 328). Although the cause of these cAMP surges is not known for certain, in general terms it is expected that a high ratio of adenylate cyclase to phosphodiesterase activities accounts for the elevation of cAMP and a low ratio of these enzyme activities accounts for the cAMP decrease (329). Although this seems reasonable, the hypothesis has not been rigorously tested. These relationships, however, have been observed in RPMI-8866 human lymphoid cells (330).In regenerating rat liver, an increase in the number of beta receptors on the hepatocyte membranes contributes to an increase in adenylate cyclase activity (331). In addition, a transient increase in the levels of calmodulin have been reported to coincide with the second cAMP increase seen in this tissue (318, 332). This could conceivably contribute to a stimulatory effect on adenylate cyclase and/or a stimulation of a phosphodiesterase (332, 333). Several investigators have demonstrated that cAMP elevation leads to an induction of phosphodiesterase (334339). Liu reported in 3T3-LI cells that treatment with dibutyryl-CAMP for 2-7 days leads to a 4- to 6-fold increase in the specificity activity of phosphodiesterase (337). Similar results were obtained in S-49 mouse lymphoma


87

cells (338) and mouse neuroblastoma cells (339). The observation that indomethacin, an inhibitor of prostaglandin synthesis, blocks the increase in cAMP and subsequent DNA synthesis in mouse lymphocytes suggests that prostaglandin, a known stimulator of adenyate cyclase, may be involved in the cAMP elevation (340). Rozengurt et al. (341)reported that prostaglandin E, causes an increase in intracellular cAMP and stimulates DNA synthesis in Swiss 3T3 cells when insulin is present. These investigators (342) have also demonstrated that partially purified porcine platelet-derived growth factor, a potent mitogen for untransformed fibroblastic cells, causes an increase in production of prostaglandins of the E series and a marked, indomethacin-sensitive elevation of cAMP in the presence of phosphodiesterase inhibitors. The origin of the signals that regulate these cAMP fluctuations are presently not known. There has been some speculation that signals that regulate the cell cycle are internal or intrinsic to the cells. There are examples of this type in nature. The intrinsic cardiac rhythmicity and electroencephalographic wave rhythms are of internal origin. Generation of cAMP may be one of these inherent signals of cAMP regulation superimposed on other inherent regulatory signals (317). In the intact organism, cells of a developing tissue do not function in isolation, and external signals, which act at cell surface or internal receptors, may affect cAMP levels. Hormones and growth factors are required for proliferation or differentiation. The cell cycle and these developmental processes may be regulated by both internal and external signals similar to circadian rhythms or other biological clock mechanisms. Rozengurt et al. have suggested (342),from their work and the work of others (343-343, that cells may secrete substances such as prostaglandins which then act on receptors of surrounding cells, as well as on their own. The original second-messenger hypothesis (3) presented a unifying theory describing the elevation of cAMP as an intracellular response generated by an external hormone signal. Therefore, cAMP is described as a “Director of Foreign Affairs” (3). That cAMP and the CAMP-dependent protein kinase are involved in cell development, which could be “internal affairs,” is not necessarily implicit in this theory (312). Although phosphorylation reactions are the only known mechanism of protein kinase action, it cannot necessarily be assumed that all the effects of cAMP are mediated by them. It should not be ruled out that the regulatory subunit or other unrecognized CAMP-binding proteins may play a role of their own (see Section IV,F.). Nevertheless, knowing the mechanism of activation of protein kinase by cAMP and the importance of this equilibrium reaction in the regulation of metabolism of cells in the quiescent, Go state, it is likely that protein kinase-mediated phosphorylation is involved in some aspects of regulation or modulation of cell cycle events. If cAMP andlor the protein kinase are involved in the regulation of developmental processes, an important question is at what site or sites do they act. Using

88

STEPHEN J. BEEBE AND JACKIE D.CORBIN

the role of cAMP in hormonal regulation as an analogy, it is likely that the CAMP-dependent protein kinase acts at transition phases; for example, it could act at sites between the end of one cell division and the DNA synthesis that precedes the next division. However, there is some controversy concerning the nature of events that comprise this transition (314, 317, 318). One hypothesis proposes that growth-arrested cells are in a distinct quiescent Go state which they must leave to reenter the GI phase (346-348). Friedman (317 ) has discussed the possible roles of cAMP at this point but whether or not protein kinase mediates these events, and if so, how, is not known. An alternative theory proposes that nonproliferating cells remain in the G, phase but have a very low probability of continuing into the S phase due to a number of factors related to requirements for DNA synthesis and mitosis (349). Pledger and co-workers (350) showed that quiescent BALB-c/3T3 cells must pass through at least two phases, which are separately regulated, before they can synthesize DNA. They suggested that these cells require a platelet-derived growth factor to become “competent,” that is to leave the Go phase and enter the cell cycle. Other factors are required for competent cells to progress through the cell cycle. These factors are present in platelet-poor plasma and are effective in competent but not in incompetent cells. This progression factor appears to be somatomedin C or other members of this family of growth factors and epidermal growth factor. Interestingly, simian virus 40 provides both competence and progression activity. Boynton and Whitfield (318) have reviewed this theory in detail and have suggested that there are at least four different types of prereplicative phases in eucaryotic cells. There has been considerable controversy concerning the role of cAMP in the cell cycle as a positive or negative regulator (314, 315, 317, 318). That cAMP is a negative regulator of proliferation is based on three general types of evidence. One is the observation that the levels of cAMP are often high in the quiescent state and then drop as the cells are stimulated to proliferate. The second is that the use of exogenous cyclic nucleotides or agents that elevate intracellular cAMP arrests many cells in the G, or G, phase of the cycle. A third type of evidence is derived from the use of cell varients that are deficient in the CAMP-dependent protein kinase (351-360). However, evidence is also available to support a stimulatory role for cAMP in the cell cycle. Although the addition of high concentrations of cAMP or cAMP analogs at critical points during the cell cycle inhibits cell proliferation in some cell types, at appropriate times cAMP or cAMP analogs stimulate proliferation in other cell types (314, 317, 318). Studies conducted by Wang et al. (340) using conconavalin A-stimulated mouse lymphocytes indicate that a rise and a fall in cAMP is required for the progression of these cells into the S phase of the cell cycle (318, 340, 361). When conconavalin A is used to stimulate lymphocytes to proliferate, RNA synthesis precedes and DNA synthesis follows the increase in CAMP. When indomethacin, which stimulates adenylate cyclase, is used to block synthesis of


89

prostaglandins, both the cAMP increase and DNA synthesis are blocked. This condition is reversed when indomethacin is removed. The addition of 8-bromocAMP or the phosphodiesterase inhibitor RO-20- 1724 after the cAMP rise, but before DNA synthesis, blocks the progression of concanavalin A-stimulated cells into the S phase. When the first peak of cAMP is prevented from falling in serum-starved human fibroblasts, DNA synthesis is not blocked (362). However, if the subsequent late GI phase cAMP surge is prevented from subsiding, DNA synthesis is arrested. Likewise, prevention of the natural rise and fall of cAMP in the late G , phase of human lymphoid cells prevents mitosis (363). It should be kept in mind that it is difficult to generalize the effects of cAMP on cell development, and the effects of the cyclic nucleotide may be tissue- and/or cell stagespecific (314, 315, 317). Other evidence not only supports a role in cell development for cAMP but also for the CAMP-dependent protein kinase isozymes (237, 31 7-320). The method generally used to quantitate the CAMP-dependent protein kinase isozymes is by the use of DEAE-cellulose chromatography. Although this technique can be used for reasonably accurate estimations of protein kinase isozymes, it is important to realize and avoid the potential pitfalls of the method. Specific antibodies for each isozyme have been used and provide a more accurate determination of isozyme levels (237). The levels of the CAMP-dependent protein kinase isozymes have been measured during the cell cycle in several studies. In these studies the levels of both isozymes are reported to oscillate during cell cycle traverse. While the oscillation of isozymes suggests that each may have a separate and specific function, there are little, if any, data to indicate what these functions might be. Costa et al. (364) observed in Chinese hamster ovary cells that the specific activity of CAMPdependent protein kinase first decreases about twofold and then increases I .5to 3.5-fold such that a maximum was reached at the G,-S border. As the cells proceed through the G , phase, the ratio of type I to type 11 protein kinase changes reciprocally. The increase in specific protein kinase activity is due to a selective elevation of the type I1 isozyme while the type 1 isozyme decreases. Haddox er al. (365) observed that as the Chinese hamster ovary cells traverse GI and approach the S phase, the type I1 isozyme increases as a function of time and cAMP concentration, but the specific activity of type I does not change throughout the cycle. However, if dibutyryl- or 8-bromo-CAMPis included in the growth medium after detachment, cell growth is arrested and a dramatic, cycloheximidesensitive increase in the type I isozyme occurs. The type 11 isozyme decreases by this treatment. Estimation of the half-lives of the isozymes indicates that the cyclic nucleotide treatment causes a selective turnover of the type I1 isozyme. Friedman and co-workers correlated the levels of cAMP with the levels and the activation state of the CAMP-dependent protein kinase in HeLa cells (366). Cells in the log phase of growth have nearly equal amounts of type I and type I1

90


isozymes but the enzyme activity ratio changes are relatively insensitive to high or changing levels of CAMP.This may be due to a higher KO for cAMP activation at high levels of protein kinase. However, during mitosis, the cAMP levels and the activity ratio decrease to their lowest levels. The protein kinase levels remain high during mitosis and then fall after mitosis. Although no separate and specific roles for the CAMP-dependent protein kinase isozymes have been clearly established, it has been suggested that the type I isozyme is a positive effector of growth and the type I1 isozyme relates more to tissue differentiation (319, 320). A number of different approaches have been used to study these phenomena. One approach has been to study the relative proportions of the CAMP-dependent protein kinase isozymes during whole-organ development. Lee et al. (367) demonstrated a rapid increase in rat testicular protein kinase activity during the first postnatal week which is correlated primarily with an increase in the type I isozyme. An increase in the type I1 isozyme occurs coincidently with the onset of complete spermatogenesis. Eppenberger et al. (368), studying postnatal uterine development, indicated that the ratio of type I to type I1 protein kinase decreases from about 0.55 on day 1 to about 0.1 on day 20, then increases to the day 1 level by day 40. These ratio changes are due to a decrease and then an increase in the type I isozyme with no change in the type 11 isozyme. Wittmaack et al. (369)used immunoprecipatation techniques in developing rat liver and malignant hepatic tissues. In developing liver, the total inhibitor-sensitive protein kinase activity steadily increases during normal liver development until the maximum levels are reached in the adult liver. The type Itype I1 ratio is about 1.2 in the fetus (4 days prepartum), 0.7 at birth, 1.8 on day 17 after birth, and 1.2 in the adult. Another approach to demonstrate a specific role for one of the isozymes in proliferation is to evaluate isozyme ratios in normal and malignant tissues from the same organ. Fossberg ef al. (370), comparing extracts of normal and carcinoma-involved human renal cortex, and Handschin and Eppenberger, studying normal and malignant human mammary tissue (371),found that the type I to type I1 ratio is approximately two times higher in extracts from malignant tissue, although the total protein kinase activity is the same in both normal and malignant tissue. Weber et al. (372) and Wittmaack et al. (369)suggest that a simple correlation of one isozyme with proliferation and the other with differentiation may not hold. They point out that in some of these systems changes in proliferation are also associated with changes in differentiation. Lymphocytes from normal patients and patients with chronic lymphocytic leukemia represent pure cell-types that show negligible proliferation and differ only in the state of differentiation (372). The tumor lymphocytes contain lower levels of CAMP, inhibitor-sensitive CAMP-dependent protein kinase activity and correspondingly lower levels of regulatory subunit. By using SDS-gel electrophoretic separation of 32P-labeled


91

8-azido-CAMPregulatory subunit and type 1- and type 11-specific immunotitration, differences in the regulatory subunit pattern are seen between the normal and leukemic lymphocytes. Approximately 60% of the total CAMP-binding sites are associated with the type I isozyme in both normal and leukemic cells. About 40% of the CAMP-binding sites are associated with the type I1 isozyme in leukemic lymphocytes, but in normal cells 20% of the sites are associated with the type I1 isozyme and about 20%are represented as low immunoreactive type I binding sites. Wittmaack et al. (369) have also demonstrated that the type I to type I1 ratio in rapidly proliferating AH- 130 hepatoma cells is higher than in the normal adult liver but does not change when the cells go into the stationary phase. In addition, the undifferentiated AH-1 30 hepatoma cells have a lower ratio than do the well-differentiated 9618-A cells. Consequently, these data do not support the type I-proliferation, type 11-differentiationhypothesis but suggest that the ratio of type I to type I1 relates more to the terminal differentiation of the organ than to the proliferation rates. Byus et al. (373) reported that proliferation of concanavalin A-stimulated human peripheral lymphocytes leads to a cAMP increase and a selective activation of the type I isozyme of the CAMP-dependent protein kinase, even though both isozymes are present. Incubation of lymphocytes with both concanavalin A and dibutyryl-CAMP causes an activation of both type I and type I1 isozymes and prevents both RNA and DNA synthesis. These data led them to conclude that the type 1 isozyme is a positive modulator of lymphocyte proliferation while activation of the type I1 isozyme, or activation of both isozymes, inhibits proliferation. The conclusion of Wang et al. (340),that there is a requirement for a rise and fall in CAMP, is not inconsistent with these data. The conconavalin A could allow the rise and fall in cAMP and subsequent cell division but the inclusion of dibutyryl-CAMP would not allow the fall in cyclic nucleotide, and proliferation could be prevented. Several lines of evidence indicate that the CAMP-dependent protein kinase isozyme ratio changes following viral or chemical transformation. Gharrett et al. (374)and Wehner et al. (375)characterized and compared the CAMP-dependent protein kinase isozymes in normal 3T3 cells with SV40-transformed (374, 375), spontaneously transformed ( 3 7 3 , and methylcholanthrene-transformed (375) cell counterparts. While normal cells contain only the type I1 isozyme, transformed cells contain an equivalent amount of kinase activity but these cells contain both type I and type I1 activities as determined by all three of these techniques. Ledinko and Chan (376) also observed that the type I isozyme is higher in rat 3Y 1 cells transformed by human adenovirus type 12 compared to untransformed 3Y1 cells. Little or no change occurs in the type I1 isozyme following transformation. However, Haddox et al. (365) determined that Rous sarcoma virus-transformed fibroblasts have a greater protein kinase specific activity and a lower ratio

92

STEPHEN J. BEEBE AND JACKIE D. CORBM

of type I to type I1 isozyme than do the normal cells (0.42 in the transformed cells and 0.85 in the normal cells). The higher amount of the type I1 isozyme in these transformed cells compared to normal cells is in contrast to the higher amount of the type I isozyme in the SV40-transformed 3T3 cells observed by Gharrett et al. (374) and Wehner et al. (375) and in the adenosine-transformed 3YI cells described by Ledinko and Chan (376). It is not clear if these differences are due to the cell type, the transforming virus, or both. Several laboratories have reported changes in CAMP-dependent protein kinase isozymes during cellular differentiation. Conti et al. (377) separated germ cells from mouse testes and determined the quantity of the two isozymes of protein kinase in preparations enriched (70-90%) in middle-late pachytene spermatocytes, round spermatids, and elongated spermatids by measuring CAMPdependent kinase activity using protamine as substrate and [3H]cAMP binding activity following DEAE-cellulose chromatography. Pachytene spermatocytes had a type I to type I1 isozyme ratio of about 2.0. Cells in later stages of spermatogenesis had lower amounts of the type 1 isozyme and increasing amounts of the type I1 isozyme. Round spermatids had a type I to type 11 isozyme ratio of about 1 .O while elongated spermatids contained almost exclusively the type 1 isozyme. Elongated spermatids appeared to contain lower total CAMPdependent protein kinase activity than do pachytene spermatocytes or rounded spermatids. Fakunding and Means (378) demonstrated similar isozyme changes during rat Sertoli cell maturation by measuring [3H]cAMP binding and inhibitor-sensitive, CAMP-dependent histone kinase activity after DEAE-cellulose chromatography. At 12-days of age, Sertoli cells contained 2.5-3.0 times more type I than type I1 isozyme. Further development was associated with a decrease in the type I to type I1 isozyme ratio such that after 20-days of age the ratio was approximately 0.8. Schwartz and Rubin (379) have used CAMP-dependent histone phosphorylation and [3H]cAMP binding following DEAE-cellulose chromatography in combination with type I and type I1 specific immunoprecipitation to study changes in the amounts of regulatory and catalytic subunits in Friend erythroleukemic cells before and after stimulation of differentiation with dimethylsulfoxide. During differentiation, the concentration of the type I1 regulatory subunit increased threefold and the type 1 regulatory subunit decreases to one-third of the control level, resulting in a change of the type I to type I1 regulatory subunit from 0.8 in the undifferentiated cells to 0.1 after induction of differentiation. Changes in the catalytic activity were proportional to changes in binding activity, indicating that the ratio of regulatory to catalytic subunit did not change. When the cells were treated with 8-bromo-CAMPand 3-isobutyl- 1-methylxanthine for two days, similar changes in the regulatory subunit occurred without proportional changes in the catalytic subunit, but differentiation did not appear.


93

Liu (337) studied changes in the CAMP-dependent isozymes during differentiation of 3T3-Ll cells from fibroblasts to adipocytes. The regulatory subunit was determined by 8-azido-CAMP photoaffinity labeling and catalytic activity was determined by CAMP-dependent histone phosphorylation. In the undifferentiated fibroblasts, the type I1 isozyme was the predominant CAMP-dependent protein kinase. After promotion of differentiation with 3-isobutyl- 1-methylxanthine, dexamethasone, and insulin, an increase in the total CAMP-dependent protein kinase activity was attributed to a 3- to 6-fold increase in the type I isozyme. This was due to an increase in both the regulatory and the catalytic subunits. Similar results were obtained with spontaneously differentiating 3T3L1 cells, demonstrating that the selective increase in the type I isozyme appeared to be directly related to adipocyte differentiation rather than to the drug and hormone treatment used to promote differentiation. Furthermore, 3T3-C2 cells, which had a lower intrinsic ability to undergo differentiation, did not show the selective type I increase when stimulated in the same manner. How cAMP and protein kinases are involved in growth and development is poorly understood. It is likely that protein kinase is integrated with primary, intrinsic regulatory signals and other second messengers such as calcium and phosphoinositides. The consequences of this latter second messenger are only beginning to be unraveled but it is pertinent to ask how it may be involved in cell development (380, 381). Calcium is particularly interesting since it appears to play a central role in many regulatory processes (380, 381). For example, both phospholipase A,, which generates prostaglandin-like precursors (382), and phospholipase C , which generates phosphoinositides and diacylglycerol (380, 381), are calcium-sensitive enzymes. Furthermore, calcium can also regulate cAMP metabolism via regulation of adenylate cyclase and phosphodiesterase (383).A circular and more complicated, but potentially tightly regulated picture, emerges when it is considered that prostaglandins activate adenylate cyclase, phosphoinositides release intracellular calcium, and diacylglycerol activates protein kinase C, a calcium-sensitive enzyme. Cyclic AMP, via protein kinase, is known to regulate calcium levels. It will be interesting to follow future developments involving the interactions of CAMP, Ca2 +,and phosphoinositides in the regulation of growth, development, and metabolism in general. F.

VARIATIONS IN THE REGULATORY SUBUNIT-CATALYTIC SUBUNIT RATIO

The ratio of regulatory to catalytic subunit in terminally differentiated tissues has been reported to be approximately 1 .O. This has been demonstrated in several species of heart (216), in several tissues from adult rabbits (90). and in the postnatal developing rat liver (369) and rat brain (384). This suggests that the two subunits are coordinately regulated. Uno et al. (385), analyzing yeast mu-

94


tants altered in cAMP metabolism, have determined that the subunits of protein kinase are not coded by the same gene locus, suggesting the possibility that the levels of regulatory and catalytic subunits could be independently regulated. There are several reports in the literature indicating that, at least under certain circumstances in some cells, there are greater amounts of regulatory subunit than catalytic subunit. Richards and Rolfes (386)studied ovarian granulosa cells from immature, hypophysectomized rats treated with estrogen and human folliclestimulating hormones. Under the influence of these hormones, granulosa cells luteinize and produce steroids. These investigators reported that the type I1 regulatory subunit is induced 10- to 20-fold in treated animals, with little if any corresponding increase in catalytic activity. Darbon et al. (387)reported a similar finding. Segaloff et al. (281) have shown that porcine granulosa cells in primary culture respond to treatment with cholera toxin, follicle-stimulating hormone, and cAMP analogs by an increased specific activity of the type 11, but not of the type I regulatory subunit. A 5 - to 10-fold increase in the regulatory subunit determined [3H]cAMP binding and densitometric scans of autoradiograms from specific 8-azido["H]cAMP photoaffinity labeling of the regulatory subunit, is correlated with a 2- to 4-fold increase in the specific catalytic subunit activity, measured by heptapeptide phosphorylation. Walter et al. (388) and Lohmann et al. (389) studied the potential role of protein kinase in the dibutyryl-CAMP-stimulated differentiation of neuroblastoma-glioma hybrid cells. Treatment of hybrid cells with dibutyryl-CAMP or addition of agents that elevate intracellular cAMP results in a selective increase in the type I regulatory subunit with no change in the catalytic subunit or type I1 regulatory subunit. The type I regulatory subunit is detected by 8-azi~ o [ ~ ~ P ] c Aphotoaffinity MP labeling (388)and by an enzyme-linked immunosorbent assay and radioimmunolabeling followed by transfer from SDS-gels to nitrocellulose (389).These experiments indicate that the hybrid cells separately regulate the levels of regulatory and catalytic subunit and also suggest that the free type I regulatory subunit could possibly be involved with the expression of some differentiating function(s). Furthermore, Morrison et al. (390)have demonstrated that there are increases in the levels of mRNA coding for the regulatory subunit. It has been suggested that this altered regulatory subunit, termed R', may be elevated due to increased synthesis and decreased degradation. It is interesting that most of the known examples of tissues that appear to show an elevation of regulatory subunit, but not of catalytic subunit, are of nervous or reproductive origin or from a mutant cell line (386,394). It will be interesting to see if this is a unique characteristic of these tissues or if other tissues also show this phenomenon. Although it appears in all of these studies that the type I1 regulatory subunit is present in excess over the catalytic subunit, it has not been ruled out that some inhibitor(s) of protein kinase, other than the regulatory subunit, is also increased


95

by the same treatment. The presence of a heat-stable protein kinase inhibitor in various tissues has been known for some time (32).Beale e t a / . (395)and Tash et al. (396) reported that a low-molecular-weight heat-stable protein kinase inhibitor is increased in the Sertoli cells of rat testes after follicle-stimulating hormone treatment. Because of potential errors in determining the levels of the protein kinase catalytic activity (389), specific antibody techniques or cDNA clones for the catalytic subunit may help resolve these uncertainties. If the regulatory subunit is present in excess compared to the catalytic subunit, the mechanism that maintains these subunits in equal proportions in normal, differentiated tissues is apparently disrupted. Although the levels of many proteins are regulated at the level of transcription, regulation can also occur at the level of translation and/or degradation of messenger RNA, and at the level of protein degradation (see Section V,G). It is therefore possible that the two subunits are regulated coordinately at one level (e.g., transcription) but regulated differently at another level (e.g., translation or degradation). These two subunits may be generally maintained in stoichiometric amounts by proteolysis of either subunit that is present in excess regardless of the mechanism of regulation of individual subunit levels. For example, the free regulatory subunit is known to be much more sensitive to proteases than is the regulatory subunit which is bound to the catalytic subunit (10). The nature of the events that determine the protein kinase subunit levels has not been determined. However, the experiments outlined in this section suggest that cAMP may be important in determining these levels. Recent results by Ratoosh et al. (397) indicate that FSH, via CAMP, may increase synthesis of rabbit granulosa cell type 11 regulatory subunit by altering the levels of mRNA for this protein. If it is assumed that the mechanism of cAMP action is through the CAMP-dependent protein kinase, it is possible that this enzyme regulates its own levels. Although either the type 1 or type I1 regulatory subunit may be bound to the particulate fraction (see Section V,B.), there may be an additional function for this subunit. Evidence has appeared to suggest such a role. The regulatory subunit may participate in the transient inhibition of phosphoprotein phosphatase(s) that dephosphoryiates various protein kinase substrates. In 1977, Gergely and Bot (398) reported that the CAMP-dependent protein kinase, in the presence of CAMP, inhibited the phosphoprotein phosphatase-catalyzed dephosphorylation of phosphorylase a, but not a 32P-labeled tetradecapeptide, suggesting a substrate-mediated effect. Khatra et a/. (399) have demonstrated that homogeneous preparations of type I1 regulatory subunits, or homogeneous type I1 CAMP-dependent protein kinase in the presence of CAMP, inhibits a highmolecular-weight phosphoprotein phosphatase from rabbit skeletal muscle. This inhibition of dephosphorylation occurs when phosphorylase, glycogen synthase (phosphorylated at sites 2 and 3) or histone is used as substrate, suggesting an

96


enzyme-directed inhibition. This inhibition occurs in vitro at concentrations of regulatory subunit that occur in vivo. Through the decrease in dephosphorylation rate, the regulatory subunit of protein kinase could possibly magnify the increases in phosphate content of proteins brought about by activation of the catalytic subunit. Thus, when tissue cAMP is elevated, the phosphoprotein phosphatase is inhibited, and when cAMP declines, as occurs in some tissues by insulin action, this enzyme is activated. It remains to be proved if this mechanism has importance in the physiological regulation of phosphoprotein phosphatase, but, if so, it suggests the possibility of a multifunctional regulatory ) in the presence of subunit. It has been reported by Constantinou et al. ( 3 9 9 ~that cAMP the phosphorylated form of the rat liver type I1 regulatory subunit possesses topoisomerase activity. There are indications to suggest that this regulatory subunit can form DNA-phospho-regulatory subunit-CAMP complexes and relax superhelixes of DNA. If this report can be confirmed and extended, a new concept in the transcriptional action of the CAMP-dependent protein kinase may emerge. Further experimentation will be required to elucidate the mechanisms that regulate the levels of protein kinase and to determine if the free regulatory subunit has a role(s) other than regulating protein kinase catalytic activity. If the regulatory subunit does have another function, the nomenclature of this enzyme as a protein kinase, which specifies a phosphorylation reaction, will be appropriate only for the phosphotransferase function. Although it has not been conclusively established that the regulatory subunit has a catalytic function, it may serve as a regulatory protein in capacities other than for protein kinase activity. Since it has some sequence homologies with the catabolite activator protein (CAP) of E . coli (156), it is tempting to speculate that it may serve as a regulator of gene transcription in mammalian cells. However, there is presently no direct evidence to support this idea, and it is known that the regulatory subunit does not contain the corresponding DNA binding domain found in CAP.

G. REGULATION OF THE AMOUNT OF CERTAIN PROTEINS BY CYCLICAMP-DEPENDENT PROTEINKINASES The postsynthetic modification of some preexisting enzymes by CAMP-dependent phosphorylation is known to alter the activities of these enzymes and thereby modify various short-term cellular functions. In recent years it has become clearer that cAMP is also involved in more long-term regulation of cellular functions through the regulation of the amounts of specific proteins. Since protein phosphorylation is a well-established mechanism of CAMPdependent protein kinase action, this mechanism may be involved in the phosphorylation of nuclear proteins, which could regulate gene transcription or other nuclear events. Although both histone and nonhistone nuclear proteins are known to be phosphorylated by the CAMP-dependent protein kinase in vitro, the


97

most convincing evidence for in vivo CAMP-mediated nuclear protein phosphorylation is for histone H I . Langan demonstrated an in vivo increase in [32P]phosphate incorporation of histone HI in the liver of rats injected with glucagon or dibutyryl-CAMP (31). To rule out artifactual phosphorylation during the extraction procedure, Wicks et al. (400) extracted histone HI with sulfuric acid and suggested a role for the CAMP-dependent protein kinase in the induction of tyrosine aminotransferase in Reuber H35 hepatoma cells. Harrison et al. (401) observed complex changes in the [32P]phosphate content of histone H1 subspecies in rat C6 glioma cells. Isoproternol elicited both increases and decreases in [32P]phosphatelevels at several phosphorylation sites. Although CAMP-independent phosphorylation apparently occurs, CAMP-dependent phosphorylation at seine-37 of the amino terminus of histone H1-1 and H1-2 in vitro correlated with increased 32Pcontent at these sites in vivo following treatment of these cells with isoproterenol. Dibutyryl-CAMP also caused complex changes in histone H 1 phosphorylation but these are different from those for isoproterenol. The physiological significance of histone phosphorylation is still not clear and has recently been reviewed by Johnson (402). Briefly, histone H1 is believed to be involved in the formation of higher-order chromatin structures and it has been suggested that phosphorylation of this protein alters the circular dichroism of histone-DNA complexes and the template activity of chromatin. Phosphorylation of serine-38 of histone HI reduces histone binding to DNA and may function to unwind tightly coiled chromatin to expose regions for RNA polymerase binding and transcription. Langan (31) has determined that more liver histone is phosphorylated at serine-38 than would be required to actively transcribe specific sequences in response to hormones or CAMP. This and other data suggest that specific base sequence recognition due to histone H1 phosphorylation is unlikely. Alternatively, Harrison et al. (401) calculated that only a very small number of glioma cell histone H 1 molecules are phosphorylated and suggest that these may occur at selected gene loci to alter chromatin function. Several reports have appeared documenting the in vitro CAMP-dependent and CAMP-independent protein kinase phosphorylation of RNA polymerase 11, the enzyme that catalyzes the transcription of genes coding for messenger RNA. Some investigators report that phosphorylation is correlated with increases in polymerase activity (403-406) while others are unable to support this finding (407-409). Kranias et al. (405) reported that nuclear c AMP-dependent protein kinase leads to incorporation of 0.5 mol [32P]phosphate/mol enzyme into the 25,000-dalton subunit of calf thymus RNA polymerase 11 with a concomitant 3fold increase in enzyme activity. Dephosphorylation by E . coli phosphatase results in a loss of 32P label and a corresponding decrease in enzyme activity. The in vivo, isoproterenol-stimulatedphosphorylation of rat C6 glioma cell RNA polymerase I1 has been reported by Lee et al. (410). Stimulation of cells results in 32P incorporation into serine residues of all six polymerase subunits with a total incorporation of 0.5-2.0 mol phosphate/mol enzyme. Propranolol inhibits

98


and dibutyryl-CAMP mimics the isoproterenol effects. The phosphorylation of RNA polymerase I, which catalyzes the transcription of ribosomal genes, and RNA polymerase 111, which transcribes genes coding for 5 S ribosomal RNA, have also been reported [see Ref. (40211. If the CAMP-stimulated induction of specific proteins requires phosphorylation event(s), the presence of the CAMP-dependent protein kinase or its catalytic subunit in the nucleus or a translocation to the nucleus would be important. Since cAMP is known to be synthesized at the plasma membrane, it seems probable that some mechanism may be required to translocate either cAMP and/or protein kinase to the nucleus. There is, however, some evidence to indicate that CAMPdependent protein kinases are in the nucleus and translocation of protein kinase to the nucleus may also occur (see Section V,B.). Agents that elevate the intracellular cAMP concentration have been shown to result in an apparent translocation of protein kinase activity from the cytoplasm to the nucleus in several tissues [see Refs. (36, 237,402, 404) for reviews]. It has been suggested that some of these earlier studies may be subject to artifacts due to enzyme redistribution and nonspecific ionic interaction of the catalytic subunit of protein kinase with subcellular binding sites when homogenization is carried out in lowionic-strength buffers (234). A number of other potential pitfalls of translocation studies have been reviewed, including cytoplasmic contamination and CAMPindependent protein kinases other than the catalytic subunit of protein kinase (10). However, nonaqueous procedures are now generally used to minimize redistribution and nonspecific binding during nuclei isolation (411). Cho-Chung (412, 413) proposed that tumor regression is dependent on translocation of a CAMP-protein kinase type I1 holoenzyme ternary complex from the cytoplasm to the nucleus. Although these studies and others (36, 237, 402, 404) suggest a translocation of protein kinase activity to the nucleus, they are not in agreement on whether it is the holoenzyme or free catalytic subunit that is translocated. Histochemical or immunocytochemical techniques have also been used to address the question of protein kinase translocation. This technique complements the purely biochemical approaches, and provides a sensitive technique to visually localize protein kinases which may be present at low concentrations in some cells or cell compartments. This approach avoids certain potential artifacts of in vitro biochemical methods and allows a more definite assignment of the presence of protein kinases in the nucleus (also see Section V,B.). Although it cannot be totally ruled out that the appearance of immunoreactive protein kinase in the nucleus is due to “uncovering” enzyme already present in the nucleus, the data are consistent with biochemical data suggesting translocation of protein kinase from the cytoplasm to the nucleus. Early experiments concerning the CAMP-stimulated induction of tyrosine aminotransferase and phosphoenolpyruvate carboxykinase suggested that regulation occurs at a posttranscriptional level, most likely involving an enhancement in the


99

rate of translation of preexisting mRNAs (414. 415). This conclusion was based on studies using inhibitors of protein synthesis and comparing the kinetics of CAMP- and glucocorticoid-stimulated protein induction. While these earlier studies determined protein induction by measuring enzyme activities, later studies quantitated specific mRNA levels by indirect in v i m translation assays and in some cases more directly by measuring mRNA levels and/or specific gene transcription, using specific cDNA probes. Results from these later studies suggested regulation at a pretranslational step and emphasized the need to reevaluate the cAMP mechanism for the specific induction of enzymes in eucaryotes. Using indirect quantitation of specific mRNA levels by in vitro translation, it has been determined that cAMP regulates the functional levels of mRNA or mRNA translational efficiency for phosphoenolpyruvate carboxykinase (416418), tyrosine aminotransferase (419, 420), lactate dehydrogenase A subunit (421), and alkaline phosphatase (422). The proportional changes in the specific mRNAs and the enzymes for which they code indicate that translation is not the primary site of regulation for these proteins. Since this approach measures translationally active mRNAs, it is not possible to determine if cAMP affects transcription, posttranscriptional modification of primary gene transcripts, increases in translatability of mRNAs, or decreased breakdown or stability of mRNAs. Experiments using specific cDNA probes indicate that cAMP regulates the abundance of mRNA for albumin in mouse hepatoma cells (423), phosphoenolpyruvate carboxykinase in rat liver (424-426), and lactate dehydrogenase A subunit in rat glioma cells (427). Evidence for transcriptional action by cAMP on induction of pituitary prolactin (428),phosphoenolpyruvate carboxykinase (429-433), lactate dehydrogenase A subunit ( 4 3 4 , and tyrosine aminotransferase (435,436) have also been reported. The CAMP-stimulation of phosphoenolpyruvate carboxykinase gene transcription has been shown to be via the CAMP-dependent protein kinase (280).The use of cAMP analog combinations to specifically cause a synergistic increase in mRNAPEPCKand gene transcription clearly indicate that these effects occur following protein kinase activation. The type 11 isozyme is the predominant isozyme and it is primarily responsible for mediating the response (see Section V,C.). Since the catalytic subunit is common to both isozymes and determines the substrate specificity, and because all known CAMP-dependent protein kinase effects are due to phosphorylation, it is likely that PEPCK gene transcription is mediated by a phosphorylation event. Results obtained from fusion of hepatoma cells with red cell ghosts loaded with either catalytic subunit or protein kinase inhibitor indicate that the catalytic subunit and not the regulatory subunit is responsible for regulation of tyrosine aminotransferase gene transcription (210). However, the finding by Constantinou et al. (399a) that a complex of phosphotype I1 regulatory subunit-CAMP may be a topoisomerase implicates the regulatory subunit in the regulation of transcription.

100


In summary, these studies generally indicate that CAMPregulates the levels of some proteins by some mechanism that involves an enhanced transcription of specific genes coding for the regulated proteins. In the case of PEPCK and tyrosine aminotransferase, the CAMP-dependent protein kinase appears to be the mediator of the transcription of these genes but the putative phosphorylated proteins responsible have not been identified. It has been reported that chromosome-associated proteins such as histones and high-mobility group proteins are phosphorylated but demonstration of associated functional alterations of these proteins has been difficult. It is reasonable to consider that phosphorylation of these proteins may play some role in gene transcription, but since specificity is in question, this may serve a general function which may facilitate another mechanism mediating specific gene transcription. A specific mechanism could possibly involve phosphorylation and activation of mRNA polymerase 11. However, to clearly demonstrate the physiological significance of this, a rigorous analysis of these reactions is required. It also remains to be clearly demonstrated whether or not the regulatory subunit plays a role in the regulation of gene transcription. ACKNOWLEDGMENTS The authors are grateful to Dr. D. Friedman, Dr. D. Granner, and Dr. R. Uhing for helpful discussions during the course of this work. We would also like to thank the many scientists who provided us with published and unpublished data during the preparation of this text. We are also very appreciative of Mrs. Penny Stelling, who spent many hours typing and editing this manuscript.

REFERENCES 1. Burnett, G., and Kennedy, E. P. (1954). JBC 211, 969. 2. Rabinowitz, M. (1962). ”The Enzymes,” 2nd ed., Vol. 6, p. 119. 3. Robinson, G. A., Butcher, R. W., and Sutherland, E. W. (1971). “Cyclic AMP.” Academic

Press, New York. 4. Sutherland, E. W. (1972). Science 177, 401. 5. Krebs, E. G. (1972). Curr. Top. Cell. Regul. 5 , 99. 6. Corbin, J. D., Sugden, P. H., West, L., Flockhart, D. A,, Lincoln, T. M., and McCarthy, D. (1978). JBC 253, 3997. 7. Corbin, J. D., Keely, S. L., and Park, C. R. (1975). JBC 250, 218. 8. Weber, W., and Hilz, H. (1979). BBRC 90, 1073. 9. Granot, J., Mildvan, A. S., Hiyama, K., Kondo, H., and Kaiser, E. T. (1980). JBC 255, 4569. 10. Flockhart, D. A;, and Corbin, J. D. (1982). CRC Crit. Rev. Biochem. 12, 133. 11. Beavo, J. A., and Mumby, M. C. (1982). Handb. Exp. Pharmakol. 581, Part I , 363. 12. D@skeland,S. O., and Qgreid, D. (1981). Int. J. Biochem. 13, 1. 13. Krebs, E. G., and Beavo, J. A. (1978). Annu. Rev. Biochem. 48, 923. 14. Nimmo, H. G., and Cohen, P. (1977). Adv. Cyclic Nucleoride Res. 8, 146. 15. Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975). Adv. Cyclic Nucleozide Res. 5 , 241. 16. Chock, P. B., and Stadtman, E. R. (1977). PNAS 74, 2766.


101

17. Goldbeter, A,, and Koshland, D. E., Jr. (1981). PNAS 78, 6840. 18. Laporte, D. C., and Koshland, D. E., Jr. (1983). Nature (London) 305, 286. 19. Koshland, D. E., Goldbeter, A., and Stock, J. B. (1982). Science 217, 220. 20. Shacter, E., Chock, P. B., and Stadtman, E. R. (1984). JBC 259, 12252. 20a. Meinke, M. H., Bishop, J. S., and Edstrom, R. D. (1986). PNAS (in press). 21. Corbin, J. D., and Deskeland, S. 0. (1983). JBC 258, 11391. 22. Gill, G. N . , Holdy, K. E., Walton, G. M., and Kanstein, C. B. (1976). PNAS 73, 3918. 23. Lincoln, T. M., and Corbin, J. D. (1977). PNAS 8, 3239. 24. MacKenzie, C. W., 111 (1982). JBC 257, 5589. 25. Lincoln, T. M., and Corbin, J. D. (1983). Adv. Cyclic Nucleotide Res. 15, 139. 26. Gill, G. N., and McCune, R. W. (1979). Curr. Top. Regul. 15, I . 27. Corbin, J. D., and Lincoln, T. M. (1978). Adv. Cyclic Nucleoride Res. 9, 159. 28. Lincoln, T. M., and Corbin, J. D. J. Cyclic Nucleoride Res. 4, 3. 29. Gill, G. N. (1977). J. Cyclic Nucleotide Res. 3, 153. 30. Walsh, D. A . , and Krebs, E. G. (1973). Enzymes 8, 555. 31. Langan, T. A. (1973). Adv. Cyclic Nucleotide Res. 3, 99. 32. Walsh, D. A., and Ashby, C. D. (1973). Recent Prog. Horm. Res. 29, 329. 33. Rubin, C. S . , and Rosen, 0. M. (1975). Annu. Rev. Biochem. 44, 831. 34. Rosen, 0. W., Erlichman, J., and Rubin, C. S. (1975). Adv. Cyclic Nucleoride Res. 5, 253. 35. Corbin, J. D., Keely, S. L., Soderling, T. R., and Park, C. R. (1 975). Adv. Cyclic Nucleoride Res. 5, 265. 36. Jungmann, R. A., and Russell, D. H. (1977). Life Sci. 20, 1787. 37. Cohen, P. (1978). Curr. Top. Cell. Regul. 14, 118. 38. Carlson, G. M., Bechtel, P. J., and Graves, D. J. (1979). Adv. Enzymol. 50, 41. 39. Glass, D. B., and Krebs, E. G. (1980). Annu. Rev. Pharmacol. Toxicol. 20, 363. 40. Robinson-Steiner, A., and Corbin, J. D. (1986). In “Handbook of Cardiology” (in press). 41. Rosen, 0. M., and Krebs, E. G . (1981). Cold Spring Harbor Conf. Cell Proliferation 8. 42. Ramsyer, J . , Kaslow, H. R., and Gill, G. N. (1974). BBRC 59, 813. 43. Dills, W. L., Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975). Adv. Cyclic Nucleoride Res. 5, 829. 44. Dills, W. L., Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975). BBRC 62, 70. 45. Rieke, E., Pauitz, N., Eigel, A,, and Wagner, K. G. (1975). Hoppe-Seyler’s Z. Physiol. Chem. 356, 1177. 46. Rannels, S. R., and Corbin, J. D. (1983). “Methods in Enzymology,” Vol. 99, p. 55. 47. Corbin, J. D., and Rannels, S. R. (1981). JBC 256,. 11671. 48. Goodwin, C. D., Beavo, J. A,, Dills, W. L., and Krebs, E. G. (1977). FP 36, 728. 49. Seville, M., and Holbrook, J. J. (1984). Anal. Biochem. 137, 330. 50. Reimann, E. M., and Beham, R. A. (1983). “Methods in Enzymology,” Vol. 99, p. 51. 51. Stralfors, P., and Belfrage, P. (1982). BBA 721, 434. 52. Muniyappa, K., Leibach, F. H., and Mendicino, J. (1983). Mol. Cell. Biochem. 50, 157. 53. Beavo, J. A,, Bechtel, P. J., and Krebs, E. G. (1974). “Methods in Enzymology,” Vol. 38C, p. 299. 54. Bechtel, P. J., Beavo, J. A., and Krebs, E. G. (1977). JBC 252, 2691. 5 5 . Sugden, P. H., Holladay, L. A , , Reimann, E. M., and Corbin, J. D. (1976). BJ 159, 409. 56. Walsh, D. A., Perkins, J. P., and Krebs, E. G. (1968). JBC 243, 3763. 57. Hofmann, F., Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975). JBC 250, 7795. 58. Hoppe, F., and Wagner, K. G. (1977). FEES Len. 74, 95. 59. Deskeland, S. 0. (1978). BBRC 83, 542. 60. Taylor, S. S., Lee, C., Swain, L., and Stafford, P. H. (1976). Anal. Biochem. 76, 45. 61. Taylor, S. S . , and Stafford, P. H. (1978). JBC 253, 2284.

102


62. Rubin, C. S., Erlichman, J., and Rosen, 0. M. (1972). JBC 247, 36. 63. Rubin, C. S., Erlichman, J . , and Rosen, 0. M. (1974). “Methods in Enzymology,” Vol. 38, p. 308. 64. Cobb, C., and Corbin, J . D. (1987). “Methods in Enzymology” (in press). 64a. Cobb, C. E., Beth, A. H., and Corbin, J. D. (1986). F.P. 45, 1800. 65. Rosen, 0. M., Rangel-Aldao, R., and Erlichman, J. (1977). Curr. Top. Cell. Regul. 12, 39. 66. Lincoln, T. M . , Dills, W. L., and Corbin, J . D. (1977). JBC 252, 4269. 67. Flockerzi, V., Speichermann, N., and Hofmann, F. (1978). JBC 253, 3395. 68. de Jonge, H. R. (1981). Adv. Cyclic Nucleoride Res. 14, 315. 69. Nakazawa, K., and Sano, M. (1975). JBC 250, 7415. 70. de Jonge, H. R., and Rosen, 0. M. (1977). JBC 252, 2780. 71. Gerzer, R., Hofmann, F., and Schultz, G. (1981). EJB 116, 479. 72. Dills, W. L., Goodwin, C. D., Lincoln, T. M., Beavo, J. A,, Bechtel, P. J., Corbin, J. D . , and Krebs, E. G. (1979). Adv. Cyclic Nucleoride Res. 10, 199. 73. Chen, L. J . , and Walsh, D. A. (1971). Biochemistry 10, 3614. 74. Yamamura, H., Inoue, Y., Shimomura, R., and Nishizuka, Y. (1972). BBRC 46, 589. 75. Kumon, A., Nishiyama, K., Yamamura, H., and Nishizuka, Y. (1972). JBC 247, 3726. 76. Yamamura, H., Nishiyama, K., Shimomura, R., and Nishizuka, Y. (1973). Biochemistry 12, 856. 77. Peters, K. A,, Demaille, J. G., and Fischer, E. H. (1977). Biochemistry 16, 5691. 78. Kubler, D., Gagelman, M., Pyerin, W., and Kinzel, V. (1979). Hoppe-Seyler’s Z. Physiol. Chem. 360, 1421. 78a. Uhler, M. D., Carmichael, D. F., Lee, D. C. Chriva, J . C., Krebs, E. G., andMcKnight, G. S. (1986). PNAS 83, 1300. 78b. Reed, J., Gagelmann, M., and Kinzel, V. (1983). ABB 222, 276. 79. Shoji, S., Ericsson. L. H., Walsh, K. A,, Fischer, E. H., and Titani, K. (1983). Biochemistry 22, 3702. 79a. Jahnsen, T., Hedin, L., Kidd, V. J., Schulz, T. 2.. and Richards, J. S. (1987). “Methods in Enzymology” (in press). 80. Titani, K., Sasagawa, T., Ericsson, L. H., Kumar, S., Smith, S. B., Krebs, E. G., and Walsh, K. A. (1984). Biochemistry 23, 4193. 81. Takio, K., Smith, S. B., Krebs, E. G., Walsh, K. A , , and Titani, K. (1982). PNAS 79,2544. 82. Erlichman, I . , Rubin, C. S., and Rosen, 0. M. (1973). JBC 248, 7607. 83. Connelly, P. A,, Hastings, T. G., and Reimann. E. M. (1986). JBC 261, 2325. 84. Hedrick, J. L., and Smith, A. J. (1968). ABB 126, 155. 85. Robinson-Steiner, A. M. Beebe, S. J., Rannels, S. R., and Corbin, J. D. (1984). JBC 259, 10596. 86. Rannels, S. R., and Corbin, J. D. (1980). J. Cyclic Nucleotide Res. 6, 203. 87. Uno, I . , Udea, T., and Greengard, P. (1977). JBC 252, 5164. 88. Nesterova, M. V., Sashchenco, L. P., Vasiliev, V. Y.,and Severin, E. S. (1975). BBA 377, 271. 89. Fleischer, N., Rosen, 0. M., and Reichlin, M. (1976). PNAS 73, 54. 90. Hofmann, F., Bechtel, P. J., and Krebs, E. G . (1977). JBC 252, 1441. 91. Rubin, C. S., Rangel-Aldao, A., Sarkar, D., Erlichman, J., and Fleischer, N. (1979). JBC 254, 5797. 92. Rapoor, C. L., Beavo, J. A , , and Steiner, A. L. (1979). JBC 254, 12427. 93. Zoller, M. J . , Kerlavage, A. R., and Taylor, S. S. (1979). JBC 254, 2408. 94. Weldon, S. L., Mumby, M. C., Beavo, J. A., and Taylor, S. S. (1983). JBC 258, 1129. 95. Corbin, J. D., Soderling, T. R., and Park, C. R. (1973). JBC 248, 1813. 96. Huang, T. S., Feramisco, J. R., Glass, D. B., and Krebs, E. G. (1979). Miami Winrer Symp. 16, 449.

3. CYCLIC NUCLEOTIDE-DEPENDENT PROTElN KINASES

103

Rangel-Aldao, R., and Rosen, 0. M. (1976). JEC 251, 3375. Rangel-Aldao, R., and Rosen, 0. M. (1977). JEC 252, 7140. Rangel-Aldao, R., Kupiec, J. W., and Rosen, 0. M. (1979). JEC 254, 2499. Scott, C. W., and Mumby, M. C. (1985). JEC 260, 2274. Hemmings, B. A., Aiten, A,, Cohen, P., Rymond, M., and Hofmann, F. (1982). EJE. 127, 473. 102. Carmichael, D. F., Geahlen, R. L., Allen, S. M., and Krebs, E. G. (1982). JEC 257, 10440. 103. Yagura, T. S . , and Miller, J. P. (1981). Biochemistry 20, 879. 103a. Goodman, M. (1981). Prog. Eiophys. Mol. Eiol. 37, 105. 104. 0greid, D., Ekanger, R., Suva, R. H., Miller, J. P., Sturm, P., Corbin, J. D., and Deskeland, S. 0. (1985). EJE 150, 219. 105. Rannels, S. R., and Corbin, J. D. (1980). JEC 255, 7085. 106. Bgreid, D., and Deskeland, S . 0. (1980). FEES Lett. 121, 340. 107. Corbin, J. D., Rannels, S. R., Flockhart, D. A,, Robinson-Steiner, A. M., Tigani, M. C., Deskeland, S. O., Suva, R., and Miller, J. P. (1982). EJB 125, 259. 108. Corbin, J. D., agreid, D., Miller, J. P., Suva, R. H., Jastorff, B., and Deskeland, S. 0. (1986). JBC 261, 12081. 109. Rannels, S. R., and Corbin, J. D. (1981). JEC 256, 7871. 110. Robinson-Steiner, A. M., and Corbin, J. D. (1982). JEC 257, 5482. 111. Robinson-Steiner, A. M., and Corbin, J. D. (1983). JEC 258, 1032. 112. Beebe, S. J . , and Corbin, J. D. (1984). Mol. Cell. Endocrinol. 36, 67. 113. Beebe, S. J . , Holloway, R., Rannels, S. R., and Corbin, J. D. (1984). JBC 259, 3539. 114. Beebe, S. J., Blackmore, P. F., Chrisman, T. D., and Corbin, J. D. (1987). “Methods in Enzymology” (in press). 114a. Beebe, S. J., Blackmore, P. F., Segaloff, D. L., Koch, S. R., Burks, D., Limbird, L. E., Granner, D. K., and Corbin, J. D. (1986). Eur. Symp. Horm. Cell Reg. loth, 1986 (in press). 115. Malkinson, A. M., Beer, D. S., Wehner, J. M., and Sheppard, J. R. (1983). EBRC 12, 214. 116. Tom-Delbauffe, D., Lognonne, J., Ohayon, R., Garaset, J., and Pavlovic-Houmae, M. (1982). EJE 125, 267. 117. Corbin, J. D.. Soderling, T. R., Sugden, P. H., Keely, S. L., and Park, C. R. (1976). In “Eukaryotic Cell Function and Growth” (J. E. Dumon, B. L. Brown, and N. 1. Marshall, eds.), p. 231. Plenum, New York. I 18. Potter, R. L., and Taylor, S. S. (1979). JBC 254, 2413. 119. Fossberg, T. M., Deskeland, S. 0.. and Ueland, P. M. (1978). AEE 189, 372. 120. Rannels, S. R., Cobb, C. E., Landiss, L. R., and Corbin, J. D. (1985). JEC 260, 3423. 120a. Reimann, E. M. (1986). Biochemistry 25, 119. 121. Sugden, P. H., and Corbin, 3. D. (1976). EJ 159, 423. 122. Erlichman, J., Rosenfeld, R., and Rosen, 0. M. (1974). JEC 249, 5000. 122a. Rangel-Aldao, R., and Rosen, 0. M. (1977). JEC 252, 7140. 123. Rosen, 0. M., and Erlichman, J. (1975). JBC 250, 7786. 124. Beavo, J. A,, Bechtel, P. J., and Krebs, E. G. (1974). PNAS 71, 3580. 125. Erlichman, J., Sarkar, D., Fleischer, N., and Rubin, C. S. (1980). JEC 255, 8179. 126. Rubin, C. S., Rangel-Aldao, R., Sarkar, D., Erlichman, J., and Fleischer, N. (1979). JEC 254, 3797. 127. Jahnsen, T., Lohmann, S. M., Walter, U., Hedin, L., and Richards, J. S . (1985). JEC 260, 15980. 128. Schwartz, D. A., and Rubin, C. R., (1985). JEC 260, 6296. 129. Hartl, F. T., and Roskoski, R. (1983). JEC 258, 3950. 130. Weldon, S. L., Mumby, M. C., and Taylor, S. S. (1985). JEC 260, 6440. 131. Takio, K., Wade, R. D., Smith, S. B., Krebs, E. G., Walsh, K. A,, and Titani, K. (1984). Biochemistry 23, 4207. 97. 98. 99. 100. 101.

104

STEPHEN I. BEEBE AND JACKIE D. CORBIN

132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146.

Gill, G. N., Walton, G. M., and Sperry, P. J . (1977). JBC 252, 6443. Hofmann, F., and Flockerzi, V. (1983). EJB 130, 599. Ai&en. A., Hemmings, B. A., and Hofmann, F. (1984). BBA 790, 219. Hofmann, F., Gensheimer, H. P., and Gobel, C. (1983). FEBS Lefr. 164, 350. Hofmann, F., Gensheimer, H. P., and Gobel, C. (1985). EJB 147, 361. Takai, Y., Yamamura, H., and Nishizuka, Y. (1974). JBC 249, 530. Grankowshi, N., Kudlicki, W., and Gasior, E. (1974). FEBS Lett. 47, 103. Becker-Ursic, D., and Davies, J. (1976). Biochemistry 15, 2289. Hixson, C. S., and Krebs, E. G. (1980). JBC 255, 2137. Sy, J., and Roselle, M. (1982). PNAS 79, 2874. Uno, I., Matusumoto, K., Adachi, K., and Ishikawa, T. (1984). JBC 259, 12508. Moreno, S., and Paseron, S. (1980). ABB 199, 321. Juliani, M. H., and Maia, J. C. C. (1979). BBA 567, 347. Trevillyan, J. M., and Pall, M. L. (1982). JBC 257, 3978. Majorfeld, I. H., Leichtling, B. H., Meligeni, J. A,, Spitz, E., and Rickenberg, H. V. (1984). JBC 259, 654. Garcia, J. L., Haro, A., and Muncio, A. M. (1983). ABB 220, 509. Rutherford, C. L., Vaughan, R. L., Cloutier, M. J., Ferris, D. K., and Brickley, D. A. (1984). 23, 461 1. Foster, J. L., Guttman, J. J . , Hall, L. M., and Rosen, 0. M. (1984). JBC 259, 13049. Vardanis, A. (1980). JBC 255, 7238. Vardanis, A. (1984). BBRC 125, 947. Riggs, A. D., Reiness, G., and Zubay G. (1971). PNAS 68, 1222. Anderson, W. B., Schneider, A. B., Perlman, R. L., and Pastan, I. (1971). JBC 246, 5929. Anderson, W. F.. Ohlendorf, D. H., Takeda, Y., and Matthews, B. (1981). Nature (London) 290, 754. Pabo, C., and Lewis, M. (1982). Nature (London) 298, 443. Weber, I. T., Takio, K., Titani, K., and Steitz, T. A. (1982). PNAS 79, 7679. Takahashi, M., Blazy, B., and Bandras, A. (1980). Biochemistry 19, 5124. Bonner, J. T., Barkley, D. S . , Hall, E. M., Konijn, T. M., Mason, J. W., O’Keefe, G., and Wolfe, P. B. (1969). Dev. Biol. 20, 72. Hutchins, B. L. M., and Frazier, W. A. (1984). JBC 259, 4379. Theibert, A., Klein, P., and Devreotes, P. N. (1984). JBC 259, 12318. Rangel-Aldao, R., Tovar, G., and de Ruiz, M. L. (1983). JBC 258, 6979. Francis, S. H., Lincoln, T. M., and Corbin, J . D. (1980). JBC 255, 620. Hamet, P., and Coquil, J. F. (1978). J. Cyclic Nucleoride Res. 4, 281. Hurwitz, R. L., Hansen, R. S . , Harrison, S. A., Martins, J. T., Mumby, M. C., and Beavo, J. A. (1984). Adv. Cyclic Nucleoride Protein Phosphorylarion Res. 16, 89. Reimann, E. M., Titani, K., Ericsson, L. H., Wade, R. D., Fischer, E. H., and Walsh, K. A. (1984). Biochemistry 23, 4185. Takio, K. T., Blumenthal, D. K., Edelman, A. M., Walsh, K. A,, Krebs, E. G., and Titani, K. (1985). Biochemistry 24, 6028. Barker, W. C., and Dayhoff, M. 0. (1982). PNAS 79, 2836. Weldon, S. L., and Taylor, S. S. (1985). JBC 260, 4203. Walsh, D. A., Ashby, C. D., Gonzalez, C., Calkins, D., Fischer, E. F., and Krebs, E. G. (1971). JBC 246, 1977. Ferraz, C., Demaille, J. G., and Fischer, E. H. (1979). Biochimie 61, 645. McPherson, J. M., Whitehouse, S . , and Walsh, D. A. (1979). Biochemistry 18, 4835. Demaille, J. G., Peters, K. A., and Fischer, E. H. (1977). Biochemistry 16, 3080. Scott, J. D., Fischer, E. H., Demaille, J. F., and Krebs, E. G. (1985). PNAS 82, 4379.

147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165.

166. 167. 168. 169. 170. 171. 172. 173.


105

174. Flockhart, D. A,, Watterson, D. M., and Corbin, I . D. (1980). JBC 255, 4435. 175. Tsuzuki, I., and Kiger, J. A , , Jr. (1978). Biochemistry 17, 2961. 176. Chan, V., Huang, L. C., Romero, G . , Giltonen, R. L., and Huang, C. (1980). Biochemistry 19, 924. 177. Builder, S. E., Beavo, I. A., and Krebs, E. J. (1980). JBC 255, 3514. 178. Armstrong, R. N., and Kaiser, E. T. (1978). Biochemistry 17, 2840. 179. Doskeland, S. O., and 0greid, D. (1984). JBC 259, 2291. 180. Swillens, S. (1983). EJB 137, 581. 181. 0greid, D., Doskeland, S. O., and Miller, J. P. (1983). JBC 258, 1041. 182. Seville, M., England, P. J., and Holbrook, J. J. (1984). BJ 217, 633. 183. Brostrom, C. O., Corbin, J. D., King, C. A., and Krebs, E. G. (1983). PNAS 68, 2444. 184. Illgreid, D., and Doskeland, S. 0. (1983). Biochemisrry 22, 1686. 185. Hartl, F. T., Roskoski, R., Jr., Rosendahl, M. S., and Leonard, N. J. (1983). Biochemistry 22, 2347. 186. Zoller, M. J., and Taylor, S. S. (1979). JBC 254, 8363. 187. Zoller, M. J., Nelson, N. C., and Taylor, S. S . (1981). JBC 256, 10837. 188. Rajinder, N. P., Bhatnager, D., and Roskoski, R., Jr. (1985). FP 44, 702 (abstr.). 189. Bramson, H. N., Thomas, N., Matsueda, R., Nelson, N. C., Taylor, S. S., and Kaiser, E. T. (1982). JBC 257, 10575. 190. Kupfer, A., Jimenez, J. S., and Shaltiel, S. (1980). BBRC 96, 77. 191. Bramson, H. N . , Kaiser, E. T., and Mildvan, A. S., (1984). CRC Crir. Rev. Biochem. 15,93. 192. Flockhart, D. A,, Freist, W., Hoppe, J., Lincoln, T. M., and Corbin, 1. D. (1984). EJB 140, 289. 193. Moll, G. W., Jr., and Kaiser, E. T., (1976). JBC 251, 3993. 194. Matsuo, M., Chang, L., Huang, C., and Villar-Palasi, C. (1978). FEES Len. 87, 77. 195. Pomerantz, A. H., Allfrey, V. G., Merrfield, R. B., and Johnson, E. M. (1977). PNAS 74, 4261. 196. Kochetkov, S. N., Bulargina, T. V., Sashchenko, L. P., and Severin, E. S. (1977). EJB 81, 111. 197. Cook, P. F., Neville, M. E., Jr., Vrana, K., Hartl, F. T., and Roskoski, R., Jr. (1982). Biochemisfry 21, 5794. 198. Whitehouse, S.. Feramisco, J. R., Casnellie, J. E., Krebs, E. G., and Walsh, D. A. (1983). JBC 258, 3693. 199. Granot, J., Mildvan, A. S., Bramson, H. N., Thomas, N., and Kaiser, E. T. (1981). Biochemistry 20, 602. 200. Reed, J., and Kinzel, V. (1984). Biochemistry 23, 968. 200a. Sowadski, J. M., Xuong, N., Anderson, D., andTaylor, S. S. (1985). J . Mol. B i d . 182,617. 201. Krebs, E. G. (1973). fnr. Congr. Ser.-Excerpta Med. 273, 17. 202. Corbin, J. D. (1983). “Methods in Enzymology,” Vol. 99, p. 227. 203. Kemp, B. E., and Clark, M. G. (1978). JBC 253, 5147. 204. Steinberg, R. A. (1984). Biochem. Acrions Horm. 11, 25. 205. Coffino, P., Bourne, H. R., Friedrich, U., Hochman, J., Insel, P.A., Lamaire, I., Melrnon, K., and Tomkins, G. M. (1975). Recent Prog. Horm. Res. 32, 669. 206. Maller, J . L., and Krebs, E. G. (1977). JBC 252, 1712. 207. Ostenieder, W., Brum, G., Herscheler, J., Trautwain, W., Flockerzi, V., and Hofmann, F. (1982). Nature (London) 298, 576. 208. Brum, G., Flockerzi, V., Hofmann, F., Osterrieder, W., and Trautwein, W. (1983). Pjlueger’s Arch. 398, 147. 209. Culpepper, J. A,, and Liu, A. Y.-C. (1981). J. Cell Biol. 88, 89. 210. Boney, C., Fink, D., Schlichter, D., Cam, K., and Wicks, W. D. (1983). JBC 258, 4911.

106


211. 212. 213. 214. 215. 216. 217.

Bkaily, G., and Sperelakis, S. (1984). Am. J . Physiol. 246, H630. Huttunen, J. K., Steinberg, D., and Mayer, S. E. (1970). PNAS 67, 290. Corbin, J. D., Beebe, S. J., and Blackmore, P. F. (1985). JBC 260, 8731. Beebe, S. J . , Redmon, J. B., Blackmore, P. F., and Corbin, J. D. (1985). JBC 260, 15781. Corbin, J. D., and Keely, S. L. (1977). JBC 252, 910. Corbin, J. D., Sugden, P. H., Lincoln, T. M., and Keely, S. L. (1977). JBC 252, 3854. Krause, E. G., Will, H., Schirpke, B., and Wollenberger, A. (1975). Adv. Cyclic Nucleotide Res. 5, 473. Hui, C., Drummond, M., and Drummond, G. 1. (1976). ABB 173, 415. Jones, L. R., Maddock, S. W., and Besch, H. R., Jr. (1980). JBC 255, 9971. Horwood, D. M., and Singhal, R. L. (1976). J . Mol. Cell. Cardiol. 8, 29. Jones, L. R., Maddock, S. W., and Hathaway, D. R. (1981). BBA 641, 242. Lamers, J. M. J., Stinis, H. T., and DeJunge, H. R. (1981). FEES Lett. 127, 139. Manalan, A. S . , and Jones, L. R. (1982). JBC 257, 10052. Kirschberger, M. A., Tada, M., and Katz, A. M. (1974). JBC 249, 6166. LaRaia, P. J . , and Morkin, E. (1974). Circ. Res. 35, 293. Entman, M. L., Kaniike, K., Goldstein, M. A,, Nelson, T. E., Bornet, E. P., Futch, T. W., and Schwartz, A. (1976). JBC 251, 3140. Kranias, E. G., Mandel, F., and Schwartz, A. (1980). BBRC 92, 1370. Kranias, E. G., Schwartz, A., and Jungmann, R. A. (1982). BBA 709, 28. Uno, I . , Ueda, T., and Greengard, P. (1976). JBC 251, 2192. Rubin, C. S . , Erlichman, J., and Rosen, 0. M. (1972). JBC 247, 6135. Menon, K. M. J. (1973). JBC 248, 494. Schoff, P. K., Forrester, I. T., Haley, B. E., and Atherton, R. W. (1982). J. Cell. Biochem. 19, 1. Horowitz, J. A., Toeg, H., and Orr, G. A. (1984). JBC 259, 832. Keely, S. L., Corbin, 1. D., and Park, C. R. (1975). PNAS 72, 1501. Singer, S. J. (1974). Adv. Immunol. 19, 1 . Kapoor, C. L., and Steiner, A. L. (1982). Handb. Exp. Pharmakol. 58, Part 1, 333. Lohmann, S. M., and Walter, U. (1984).Adv. Cyclic Nucleotide Protein Phosphorylarion Res. 18, 63. Walter, U., DeCamilli, P., Lohmann, S. M., Miller, P., and Greengard, P. (1981). Cold Spring Harbor Conf. Cell Proliferation 8, 141. Steiner, A. L., Koide, Y., Earp, H., Bechtel, P. J., and Beavo, J. (1978). Adv. Cyclic Nucleotide Res. 9, 691. Harper, J. F., Wallace, R. W., Cheung, C. Y., and Steiner, A. L. (1981). Adv. Cyclic Nucleoride Res. 14, 58 1. Kapoor, C . L., and Cho-Chung, Y. S. (1983). Cancer Res. 43, 295. Van Sande, J., Huang, H. L., Steiner, A., and Dumont, J. E. (1983). CellBiol. Inr. Rep. 7, 981. Jungman, R. A,, Kuettel, M. R., Squinto, S. P., Kwast-Welfeld, J. (1987). “Methods in Enzymology” (in press). Kuettel, M. R., Schwoch, G., and Jungmann, R. A. (1984). Cell Biol. Int. Rep. 8, 949. Squinto, S. P., Kelley-Geraghty, D. C., and Jungmann, R. A. (1985). J. Cyclic Nucleotide Protein Phosphorylation Res. 10, 65. Fletcher, W. H., and Byus, C. V. (1982). J. Cell Biol. 93, 719. Byus, C. V., and Fletcher, C. H. (1982). J. Cell Biol. 93, 727. Murtaugh, M. P., Steiner, A. L., and Davies, P. J. A. (1982). J . Cell Biol. 95, 64. Vallee, R. B., DiBartolomeis, M. J., and Theurkauf, W. E. (1981). J. Cell Biol. 90, 568. Theurkauf, W. E., and Vallee, R. B. (1982). JBC 257, 3284.

218. 219. 220. 221. 222. 223. 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. 244. 245. 246. 247. 248. 249. 250.


I07

251. Miller, P., Walter, U . , Theurkauf, W. E., Vallee, R. B., and DeCamilli, P. (1982). PNAS 79, 5562. 252. Lohmann, S. M., DeCamilli, P., Einig, I., and Walter, U. (1984). PNAS 81, 6723. 253. Kapoor, C. L., and Cho-Chung, Y. S. (1983). Cell B i d . Int. Rep. 7, 49. 254. Hathaway, D. R., Adelstein, R. S., and Klee, C. B. (1981). JBC 256, 8183. 255. Stewart, A. A., Ingebritsen, T. S., Manallan, A., Klee, C. B., and Cohen, P. (1982). FEES Lert. 137, 80. 256. Sarkar, D., Erlichman, J., and Rubin, C. S. (1984). JBC 259, 9840. 257. Kuo, J. K., and Greengard, P. (1970). JBC 245, 2493. 258. Kuo, J . K. (1974). PNAS 71, 4037. 259. Walter, U . , Uno, I., Liu, A. Y.-C., and Greengard, P. (1977). JBC 252, 6494. 260. Walter, U., DeCamilli, P., Lohmann, S . M., Miller, P., and Greengard, P. (1981). Cold Spring Harbor Conf. Cell Proliferation 8, 141. 261. Lincoln, T. M., Hall, C. L., Park, C. R., and Corbin, J. D. (1976). PNAS 73, 2559. 262. Lohmann, S. M., Walter, U., Miller, P. E., Greengard, P., and DeCamilli, P. (1981). PNAS 78, 653. 262a. Tremblay, J . , Gerzer, R., Vinay, P., Pang, S. C. Beliveau, R., and Hamet, P. (1985). FEES Lett. 181, 17. 262b. Fiscus, R. R., Robles, B. T., Rapoport, R. M., Waldman, S. A,, andMurad, F. (1984). Fed. Proc. 44, 698. 263. Miyamoto, E., Petzold, G. L., Harris, J. S., and Greengard, P. (1971). BBRC 44, 305. 264. Tao, M. (1972). BBRC 46,56. 265. Hofmann, F. (1980). JBC 255, 1559. 266. Buxton, I. L. O., and Brunton, L. L. (1983). JBC 258, 10233. 267. Keely, S. L. (1977). Res. Commun. Chem. Pathol. Pharmacol. 18, 283. 268. Hayes, J . S . , Brunton, L. L., Brown, J. H., Reese, J. B., and Mayer, S. E. (1979). PNAS 76, 1570. 269. Hayes, J . S . , Brunton, L. L., and Mayer, S. E. (1980). JBC 255, 51 13. 270. Schwoch, G. (1978). BJ 170, 469. 271. Byus, C. V., Hayes, J. S., Brendel, K., and Russell, D. H. (1979). Mol. Pharmacol. 19,941. 272. Hunzicker-Dunn, M. (1981). JBC 256, 12185. 273. Livesey, S. A,, Kemp, B. E., Re, C. A,, Partridge, N. C., and Martin, T. J. (1982). JBC 257, 14983. 274. Hg, K. W . , Livesey, S. A., Larkins, R. G., and Martin, T. J. (1983). Can. Res. 43, 794. 275. Livesey, S. A , , Collier, G., Zajac, I. D., Kemp, B. E., and Martin, T. S. (1984). BJ 224, 361. 276. Mizuno, T., Itzuka, K., Ikarashi, A,, and Nohara, H. (1982). Arch. Oral B i d 27, 589. 277. Chew, C. S. (1985). JBC 260, 7540. 278. Litvin, Y., PasMantier, R., Fleischer, N., and Erlichmann, J. (1984). JBC 259, 10296. 279. Ekanger, R., Sand, T. E., Ogreid, D., Christoffersen, T., and D~iskeland,S. 0. (1985). JBC 260, 3393. 280. Beebe, S. J., Koch, S. R., Granner, D. K., and Corbin, J. D. (1985). Inr. Congr. Biochem., 13th, 649. 281. Segaloff, D. L., Beebe, S. J., Burks, D., Corbin, I. D., and Limbird, L. E. (1985). Endocrinology (Baltimore) 116, Suppl, 79. 282. Chenington, A. D., Assimacopoulos, F. D., Harper, S. C., Corbin, J. D., Park, C. R., and Exton, J. H. (1976). JBC 251, 5209. 283. Wicks, W.. (1974). Adv. Cyclic Nucleotide Res. 4, 335. 284. Postern&, T., Sutherland, E. W., and Henion, W. F. (1962). BBA 65, 558. 285. Meyer, R. B., Jr., and Miller, J. P. (1974). Life Sci. 14, 1019.

108


286. Meyer, R. B., Jr., Uno, H., Robins, R. K., Simon, L. N., and Miller, J. P. (1975).Biochemisrry 14, 3315. 287. Jastorff, B. (1976). In “Eucaryotic Cell Function and Growth” (J. E. Dumont, B. L. Brown, and M. J. Marshall, eds.), p. 379. Plenum, New York. 288. Yagura, T. S., Sigman, C. C., S t u m , P. A., Reiut, E. J . , Johnson, H. L., and Miller, J. P. (1980). BBRC 92, 463. 289. Miller, J. P. (1981). Adv. Cyclic Nucleoride Res. 14, 335. 290. Miller, J. P., Uno, H., Christensen, L. F., Robins, R. K., and Meyers, R. B., Jr. (1981). Biochem. Pharmacol. 30,509. 291. Jastorff, B. (1982). In “Cell Regulation by Intracellular Signals” (S. Swillens and J. E. Dumont, eds.), p. 195. Plenum, New York. 292. Coulsen, R., Baraniak, J., Stec, W. J., and Jastorff, B. (1983). Life Sci. 32, 1489. 293. Jastorff, B., Hoppe, J., and Morr, M. (1979). EJB 101, 555. 294. DeWit, R. J. W., Hoppe, J., Stec, W. J., Baraniak, J., and Jastorff, B. (1982). EJB 122,95. 295. Free, C. A., Chasin, M., Paik, V. S., and Hers, S. M. (1971). Biochemistry, 10 3785. 296. dechrombrugghe, B., Perlman, R., Vormus, H., and Pastan, J. (1969). JBC 244, 5828. 297. Anderson, W. B., Perlman, R. L., and Pastan, L. (1972). JBC 247, 2717. 298. McKay, D. B., Weber, J. T., and Steitz, T. A. (1982). JBC 257, 9518. 299. Scholubbers, H. G.,Van Knippenberg, P. H., Baraniak, J., Stec, W. J., Morr, M., and Jastorff, B. (1984). EJB 138, 101. 300. O’Brian, C. A., Roczniak, S. O., Bramson, H. N., Baraniak, J., Stec, W. J . , and Kaiser, E. T. (1982). Biochemistry 21, 4371. 301. Rothemel, J. D., Stec, W. J., Baraniak, J., Jastorff, B., and Parker Botelho, L. H. (1983). JBC 258, 12125. 302. Rothemel, J. D., Jastorff, B., and Parker Botelho, L. H. (1984). JBC 259, 8151. 303. VanHaastert, P. J. M., Van Driel, R., Jastorff, B., Baraniak, J., Stec, W. J., and deWit, R. J. W. (1984). JBC 259, 10020. 304. Rothemel, J. D., Perillo, N. L., Marks, J. S . , and Parker Botelho, L. H. (1984). JBC 259, 15294. 305. DeWit, R. J . W., Hekstra, D., Jastorff, B., Stec, W. J., Baraniak, J., Van Driel, R., and Van Haastert, P. J. M. (1984). EJB 142, 225. 306. Dragland-Meserve, C. J., Rothemel, J. D., Houlihan, M. J., and Botelho, L. H. P. (1985).J. Cyclic Nucleoride Prorein Phosphoryl. Res. 10, 371. 307. Corbin, J. D., Beebe, S. J., Blackmore, P. F., Redmon, J . B., Sheorain, V. S . , and Gettys, T. W . (1986). In “Femstrom Symposium on Insulin Action” (in press). 308. Abell, C. W., and Monahan, T. W. (1973). J . Cell Biol. 59, 549. 309. Pastan, I . , and Johnson, G. S. (1974). Adv. Cancer Res. 19, 303. 310. Ryan, W. L., and Heidrick, M. L. (1974). Adv. Cyclic Nucleotide Res. 4, 81. 311. MacManus, J. P., Whitfield, J. F., Boynton, A. L., and Rixon, R. H. (1975). Adv. Cyclic Nucleoride Res. 5 , 719. 312. Chlapowski, F. S . , Kelley, L. A., and Butcher, R. W. (1975). Adv. CyclicNucleotide Res. 6 , 245. 313. Pastan, I., Johnson, G. S., and Anderson, N. B. (1975). Annu. Rev. Biochem. 44, 491. 314. Friedman, D. L. (1971). Physiol. Rev. 56, 652. 315. Friedman, D. L., Johnson, R. A., and Zeilig, C. E. (1976).Adv. Cyclic Nucleoride Res. 7,69. 316. Rebhun, L. I. (1977). fnr. Rev. Cyrol. 49, 1. 317. Friedman, D. L. (1982). Handb. Exp. Pharmakol. 58, Part 2 , 151. 318. Boynton, A. L., and Whitfield, J. F. (1983). Adv. Cyclic Nucleoride Res. 15, 193. 319. Jungman, R. A., and Russell, D. H. (1977). Life Sci. 20, 1787. 320. Russell, D. H. (1978). Adv. Cyclic Nucleoride Res. 9, 493.


109

321. Burger, M. M., Bombik, B. M., Breckenridge, B., and Sheppard, J. R. (1972). Nature (London) New Biol. 239, 161. 322. Seifert, W., and Rudland, P. S. (1974). PNAS 71, 4920. 323. Russell, D. H., and Stambrook, P. J. (1975). PNAS 72, 1482. 324. MacManus. J. P., Franks, D. J., Youdale, T., and Braceland, B. M. (1972). BBRC 49, 1201. 325. Thrower, S., and Ord, M. G. (1974). BJ 144, 361. 326. Short, J., Tsukada, K., Rudert, W. A,, and Leiberman, 1. (1975). JBC 250, 3602. 327. Tsang, B. K., Rixon, R. H., and Whitfield, J. F. (1980). J. Cell. Physiol. 102, 19. 328. MacManus, J. P., Braceland, B. M., Youdale, T., and Whitfield, J. F. (1973). J. Cell. Physiol. 82, 15. 329. Ebina, Y., Iwai, H., Fukui, N., Ohtsuka, H., and Miura, Y.(1975). J. Biochem. (Tokyo) 77, 641.

330. 331. 332. 333. 334. 335. 336. 337. 338. 339. 340. 341. 342. 343. 344. 345. 346. 347. 348. 349. 350. 351. 352. 353. 354. 355. 356. 357. 358. 359. 360. 361.

Millis, A. J. T., Forrest, G. A,, and Pious, D. A. (1974). Exp. Cell Res. 83, 335. Bronstad, G., and Christoffersen, T. (1980). FEES Lerr. 120, 89. Boynton, A. L., and Whitfield, J. P. (1981). Adv. Cyclic Nucleoride Res. 14, 411. Klee, C. B., Crouch, T. H., and Richman, P. G. (1980). Annu. Rev. Biochem. 49, 489. Manganiello, V., and Vaughan, M. (1972). /“AS 69, 269. D’Armiento, M., Johnson, G. S., and Pastan, I. (1972). /“AS 69, 459. Russell, T. R., and Pastan, 1. (1974). JBC 249, 7764. Liu, A. Y. C. (1982). JBC 257, 298. Bovine, H. R., Tomkins, G. M., and Dion, S. (1973). Science 181, 952. Prasad, K. N., Sahu, S. K., and Sinha, P. K. (1976). J. Natl. Cancer Insr. (U.S.) 57, 619. Wang, T., Sheppard, J. R., and Foker, J. E. (1978). Science 201, 255. Rozengurt, E., Collins, M. K. L. and Keechan, M. (1983). J. Cell. Physiol. 116, 379. Rozengurt, E., Stroobant, P., Waterfield, M. D., Deuel, T. F., and Kechan, M. (1983). Cell 34, 265. Hammarstrom, S. (1977). EJB 74, 7. Levine, L. (1981). Adv. Cancer Res. 35, 49. Hammarstrom, S. (1982). ABB 214, 431. Levine, E. M., Becker, Y., Boone, C. W., and Eagle, H. (1965). PNAS 53, 350. Salas, J., and Green, H. (1971). Narure (London), New Biol. 229, 165. Becker, H., and Stanners, C. P. (1972).J. Cell. Physiol. 80, 51. Smith, J. A , , and Martin, L. (1973). PNAS 70, 1263. Pledger. W. J., Stiles, C. D., Antoniades, H. N., and Scher, C. D. (1979). PNAS 74,4481. Daniel, V., Litwack, G., and Tomkins, G. (1973). PNAS 70, 76. Daniel, V., Bourne, H. R., and Tomkins, G. M. (1973). Nufirre (London) New B i d . 244, 167. Insel, P. A., Bourne, H. R., Coffino, P., and Tomkins, G. M. (1975). Science 190, 896. Bourne, H. R., Coffino, P., Melmon, K. L., Tornkins, G. M., and Weinstein, Y. (1975).Adv. Cyclic Nucleoride Res. 5 , 77 I . Coffino, P., Bourne, H. R., Friedrich, U., Hochman. J., Insel, P. A,, Lemaire, I., Melmon, K. L., and Tomkins, G. M. (1976). Recent f r o g . Horm. Res. 32, 669. Steinberg, R. A., O’Farrell, P. H., Friedrich, U., and Coffino, P. (1977). Cell 10, 381. Gottesman, M. M. (1980). Cell 22, 329. Gottesman, M. (1983). “Methods in Enzymology,” Vol. 97, p. 197. Gottesman, M. M., Roth, C., Leutschuh, M., Richert, N., and Pastan, 1. (1984). J. Cell. Biochem. Steinberg, R. A. (1984). Biochem. Actions Hormo. 9, 25. Whitfield, J. F., MacManus, G. P., Rixon, R. H., Boynton, A. L., and Yondale, T. (1976).In Virro 12, I .

110


362. Rechler, M. M., Bruni, C. B., Podskalny, J. M., Warner, W., and Carchman, R. A. (1977). Exp. Cell Res. 104, 41 1. 363. Millis, A. J . T., Forrest, G. A,, and Pious, D. A. (1974). Exp. Cell Res. 83, 335. 364. Costa, M., Gerner, E. W., and Russell, D. H. (1976). JEC 251, 3313. 365. Haddox, M. K., Magun, B. E., and Russell, D. H. (1980). PNAS 77, 3445. 366. Gray, J. P., Johnson, R. A,, and Friedman, D. L. (1980). AEE 202, 259. 367. Lee, P. C., Radlo, H. D., Schweppe, J. S . . and Jungman. R. A. (1976). JEC 251, 914. 368. Eppenberger, U., Roos, W., Fabbro, D., Sury, A,, Weber, I., Bechtel, E., Huber, P., and Jungmann, R. A. (1979). EJB 98, 253. 369. Wittmaack, F. M., Weber, W., and Hilz, H. (1983). EJE 129, 669. 370. Fossberg, T. M., Doskeland, S . O., and Ueland, P. M. (1978). AEE 189, 372. 371. Handschin, J . C., and Eppenberger, U. (1979). FEES Lett. 106, 301. 372. Weber, W., Schwoch, G., Wideckens, K., Garteman, A , , and Hilz, H. (1981). EJE 120,585. 373. Byus, C. V., Klimpel, G. R., Lucas, D. O., and Russell, D. H. (1977). Nature (London) 268, 63. 374. Gharrett, A. J., Mackinson, A. M., and Sheppard, J. R. (1976). Nature (London) 264, 673. 375. Wehner, J . M., Malkinson, A. M., Wiser, M. F., and Sheppard, J. R. (1981).J. Cell. Physiol. 108, 175. 376. Ledinko, N., and Chan, D. (1984). Cancer Res. 44, 2622. 377. Conti, M., Geremia, R., and Monesi, V. (1974). Mol. Cell. Endocrinol. 13, 137. 378. Fakunding, J . L., and Means, A. R. (1977). Endocrology (Baltimore) 101, 1358. 379. Schwartz, D. A,, and Rubin, C. S. (1983). JEC 258, 777. 380. Berridge, M. J. (1984). Nature (London) 312, 315. 381. Berridge, M. J. (1984). EJ 220, 345. 382. L a m e , J., Dorian, B., Daret, D., Demond-Henri, J., and Bricaud, H. (1984). Adv. Cyclic Nucleotide Protein Phosphorylation Res. 17, 585. 383. Cheung, W. Y . , and Storm, D. R. (1982). Handb. Exp. Pharmakol. 58, Part 1, 301. 384. Lohmann, S. M., Walter, U., and Greengard, P., (1978). J . Cyclic Nucleotide Res. 4, 445. 385. Uno, I . , Matsumoto, K., Adachi, K., and Ishikawa, T. (1984). JEC 259, 12508. 386. Richards, J. S., and Rolfes, A. I. (1980). JEC 255, 5481. 387. Darbon, J . M., Knect, M., Ranta, T., Dufau, M. L., and Catt, K. J . (1984). JEC 259, 14778. 388. Walter, U., Costa, M. R. C., Breakefield, X. O., and Greengard, (1979). PNAS 76, 3251. 389. Lohmann, S. M., Schwoch, G., Reiser, G., Port, R., and Walter, U. (1983). EmboJ. 2, 153. 390. Morrison, M. R., Pardue, S., Prashad, N., Croall, D. E., and Brodeur, R. (1980). EJE 106, 463. 391. Prashad, N., Rosenberg, R. N., Wischmeyer, G., Ulrich, C., and Sparkman, D. (1979). Biochemistry 18, 2717. 392. Liu, A. Y. C., Chan, T., and Chen, K. Y. (1981). Can. Res. 41, 4579. 393. Singh, T. J . , Roth, C., Gottesman, M. M., and Pastan, I. H. (1981). JEC 256, 926. 394. Steinberg, R. A., VanDaalen Wetters, T., and Coffino, P. (1978). Cell 15, 1351. 395. Beale, E. G., Dedman, J . R., and Means, A. R. (1977). JEC 252, 6322. 396. Tash, J . S . , Dedman, J. R., and Means, A. R. (1979). JEC 254, 1241. 397. Ratoosh, S. L., Jahnsen, T., and Richards, J. S. (1985). Endocrinology (Baltimore) 116, Suppl., 79. 398. Gergely, P., and Bot, G (1977). FEES Lett. 82,269. 399. Khatra, B. S., Printz, R., Cobb, C. E., and Corbin, J. D. (1985). EERC 130, 567. 399a. Constantinou, A. I . , Squinto, S. P., and Jungmann, R. A. (1985). Cell 42, 429. 400. Wicks, W. D., Koontz, J., and Wagner, K. (1975). J. Cyclic Nucleotide Res. 1, 49. 401. Harrison, J . J., Schwoch, G., Schweppe, J. S . , and Jungmann, R. A. (1982).JEC257, 13602. 402. Johnson, E. M. (1982). Handb. Exp. Pharmakol. 58, Part 1, 507.

3. CYCLIC NUCLEOTIDE-DEPENDENT PROTEIN KINASES 403. 404. 405. 406. 407. 408. 409. 410. 41 1. 412. 413. 414. 415. 416. 417. 418. 419. 420. 421. 422. 423. 424. 425. 426. 427. 428. 429. 430. 431. 432. 433. 434. 435. 436.

111

Dahmas, M. E. (1976). Biochemistry 15, 1821. Jungmann, R . A., and Kranias, E. G. (1977). Int. J. Biochem. 8, 819. Kranias, E. G., Schweppe, J. S., and Jungmann, R. A. (1977). JBC 252, 6750. Duccman, B. W., Rose, K. M., and Jacob, S. T. (1981). JBC 256, 10755. Bell, G. I . , Valenzuela, P., and Rutter, W. J. (1977). JBC 252, 3082. Dahmas, M. E. (1981). JBC 256, 3332. Breant, B., Buhler, J. M., Sentenac, A,, and Fromageot, P. (1983). EJB 130, 247. Lee, S.-K., Schweppe, J. S., and Jungmann, R. A. (1984). JBC 259, 14695. Spielrogal, A. M., Mednicks, M. I., Eppenberger, U.,and Jungmann, R. A. (1977). EJB 73, 199. Cho-Chung, Y. S. (1980). Adv. Cyclic Nucleotide Res. 12, 1 1 I . Cho-Chung, Y. S. (1980). J. Cyclic NucleotideRes. 6, 163. Wicks, W . D. (1971). JBC246, 217. Wicks, W. D., Bamett, C. A,, and McKibben, J. B. (1974). FB 33, 1105. lynedjian, P. B., and Hanson, R. W. (1977). JBC 252, 655. Nelson, K., Cimbala, M., and Hanson, R . (1980). JBC 255, 8509. Beale, E., Katzen, C., and Granner, D. (1981). Biochemistry 20, 4878. Ernest, M. J., and Feigelson, P. (1978). JBC 253, 319. Noguchi, T., Diesterhof, M., and Granner, D. (1978). JBC 253, 1332. Derda, D. F.. Miles, M. F., Schweppe, J. S . , and Johnson, R . A. (1980). JBC 255, 1 I 1 12. Firestone, G. L., and Heath, E. C. (1981). JBC 256, 1396. Brown, P. C., and Papaconstantinou, J. (1979). JBC 254, 9397. Yoo-Warren, H., Cirnbala, M., Felz, K., Monahan, J., Leis, J., and Hanson, R. (1981). JBC 256, 10224. Beale, E., Hartley, J., and Granner, D. (1982). JBC 257, 2022. Beale, E., Andreone. T., Koch, S., Granner, M., and Granner, D. (1984). Diabetes 33, 328. Miles, M. F., Hung, P., and Jungmann, R. A. (1981). JBC 256, 12445. Maurer, R . A. (1981). Nature (London) 294, 94. Cimbala, M. A,. Larners, W. H., Nelson, K., Monahan, J. E., Yoo-Warren, H., and Hanson, R. W. (1982). JBC 257, 7629. Lamers, W., Hanson, R., and Meisner, H. (1982). PNAS 79, 5137. Granner, D., Andreone, T., Sasaki, K., and Beale, E. (1983) Nature (London) 305, 549. Sasaki, K., Cripe, T. P., Koch, S. R., Andreone, T. L., Peterson, D. D., Beale, E. G., and Granner, D. K. (1984). JBC 259, 15242. Hod, Y., Moms, S. M., and Hanson, R. W. (1984). JBC 259, 15603. Jungmann, R. A,, Kelly, D. C., Miles, M. F., and Milkowski, D. M. (1983). JBC258, 5312. Culpepper, J. A., and Lee, A. Y. (1983). JBC 258, 3812. Hashimoto, S., Schmid, W., and Schutz, G. (1984). PNAS 81, 6637.


Calmodulin-Dependent Protein Kinases JAMES T. STULL MARY H. NUNNALLY CAROLYN H. MICHNOFF Department of Pharmacology and Moss Heart Center Universiry of Texas Health Science Center at Dallas Dallas, Texas 75235

I. Introduction .................................................... A. CaZ+ as a Second Messenger . . . . . . B. Calmodulin ............................. 11. Myosin Light Chain Kinases . . . . . . . . . . . . . . . . . . . . . . . . . . , .. .. . . . . . . . . A. Introduction . . . . . . ........ .................... B. Physicochemical Properties ..................................... C. Ca2+ and Calmodulin Activation . . . D. Catalytic Properties . . . . . . . . . . . . . . . . . . E. Phosphorylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Biological Significance of Myosin Phosphorylation . . . . . . . . . . . . . . . . . 111. Multifunctional Calmodulin-Dependent Protein Kinases A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .................... B . Physicochemical Properties . . . . . . . . . . . ..... C. Ca*+ and Calmodulin Activation . . . . . . . . . . . . . D. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Phosphorylation .......... .................... F. Biological Signi Dependent Protein Kinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . ........ ....................

114 114 116 118 118 119 125 130 137 140 142 142 142 149 150 157 158 159

113 THE ENZYMES. Vol. XVII Copyright 0 1986 by Academic Press. Inc. All nghts of reproduction in any form resewed.

114

J. T. STULL, M . H . NUNNALLY, AND C. H . MICHNOFF

I. Introduction A. C A ~ AS +

A

SECONDMESSENGER

In 1883, Ringer (1-3) found that extracellular Ca2+ was required to maintain contractile activity in isolated frog hearts. Locke subsequently reported that extracellular Ca2 was required for transmission of impulses from nerve to muscle (4). These fundamental observations were the first to implicate Ca2 as an important regulator of biological processes. During the 30-40 years following these pioneering experiments, a number of physiologists demonstrated that the absence of extracellular Ca2+ affected a wide variety of cellular processes ( 5 , 6). The general biological importance of Ca2+ was extended to a role as an intracellular regulator by the observation in 1942 that Ca2 activated myosin ATPase activity (7, 8). In 1947, Heilbrunn and Wiercinski (9) showed that the injection of Ca2 into muscle fibers caused contraction. These biochemical and physiological observations stimulated research on the mechanisms by which Ca2+ could stimulate muscle contraction. Ebashi and his colleagues (10, 11) found that native tropomyosin, consisting of troponin and tropomyosin, was the Ca2+-receptor complex on thin filaments in striated muscle. They proposed that Ca2 binding to troponin causes a conformational change in F-actin transmitted via tropomyosin that results in stimulation of actin and myosin interactions. About the same time, Meyer et al. (12) proposed that Ca2+ may be required for phosphorylation of phosphorylase by phosphorylase kinase. This suggestion was subsequently confirmed by Ozawa et al. (13) who showed that Ca2+ concentrations required for phosphorylase kinase activity were similar to the Ca2+ concentrations required for activation of contractile elements in skeletal muscle. Thus, Ca2+ was identified as a common link between glycogenolysis and contraction. The identification of troponin and phosphorylase kinase as intracellular receptor proteins for Ca2 provided important biochemical models for subsequent investigations on many Ca2 -dependent processes such as cell motility, secretion, cell division, metabolism, and membrane permeability. The second messenger concept proposed by Sutherland and co-workers (14) to describe the effects of cAMP on a wide variety of cellular responses was extended to Ca2+ (15). It is now generally appreciated that Ca2+ exists in low concentrations in the cytoplasm of all resting, nonactivated cells. A variety of extracellular stimuli appropriate to a particular cell results in an increase in cytoplasmic Ca2+ concentrations and initiation of Ca2 -dependent responses. The two second messengers, cAMP and Ca2+, may both affect a cellular response (i.e., cell function may be simultaneously regulated by one or both of these second-messenger systems) (15-17). +

+

+

+

+

+

+

+

115

4. CALMODULIN-DEPENDENT PROTEIN KINASES

The compelling evidence that Ca2 was a second messenger led to investigations to identify mechanisms for Ca2 selectivity to biological responses. In particular, proteins that detect and respond to a Ca2+ signal have been identified (5, 6). These Ca2+ binding proteins include calmodulin, troponin C, parvalbumin, intestinal Ca2+-binding protein, S 100 protein, and regulatory light chain of myosins. Calmodulin is unique because it is found in most, if not all, eukaryotic cells and mediates Ca2+ control of a large number of enzymes. These two general properties confer to calmodulin the role of mediator of many Ca2 dependent cellular processes. Calmodulin is known to activate distinct protein kinases, enzymes that catalyze the transfer of the terminal phosphate of ATP to serine or threonine residues in protein substrates. The introduction of a phosphate moiety into a protein may result in marked changes in its biochemical properties and thereby provide a means for regulating particular biological processes. In this chapter, we present information on the properties of two types of Ca2+ and calmodulin-dependent protein kinases, myosin light chain kinase and a multifunctional calmodulindependent protein kinase. We use the term calmodulin-dependent to mean that an enzyme or biological process is dependent upon both Ca2 and calmodulin for activity unless noted otherwise. Phosphorylase kinase, another Ca2 +-dependent protein kinase, is discussed in Chapter 10. The information presented in this chapter relates, in particular, to the chapters entitled “Regulation of Contractile Activity” and “Phosphorylation of Brain Proteins” in Volume XVIII, Chapters 13 and 9. The name myosin light chain kinase is used to identify the enzyme originally described as a protein kinase that catalyzes the calmodulin-dependent phosphorylation of a single serine on the regulatory or phosphorylatable light chains of myosins. Although this protein kinase has also been referred to as myosin kinase, we feel this terminology should be avoided because of the possible confusion with enzymes that phosphorylate myosin heavy chains (18-20). Various names have also been used to identify other calmodulin-dependent protein kinases. A type of enzyme that phosphorylates many different proteins has been purified from many tissues. We refer to this enzyme as the multifunctional calmodulin-dependent protein kinase, but the reader should be aware that some other aliases include calmodulin-dependent glycogen synthase kinase, calmodulin-dependent protein kinase 11, tubulin kinase, and calmodulin-dependent multiprotein kinase. Since we have not contributed to investigations on these protein kinases, we will not be presumptuous in recommending that multifunctional calmodulin-dependent protein kinase is necessarily the most appropriate name. However, for the purpose of this chapter, it is used to emphasize, in contrast to myosin light chain kinase, the broad specificity of this enzyme in regards to protein substrates. This chapter also includes pertinent information on +

+

+

+

J. T. STULL, M. H. NUNNALLY, AND C. H. MICHNOFF

116

calmodulin that is not historically comprehensive, and we have relied upon reviews by other authors for the citation of certain developments and perspectives.

B . CALMODULIN 1. Physicochemical Properties Calmodulin has been purified and characterized from many types of cells, including plants and protozoa as well as vertebrate tissues (21-24). The distribution of calmodulin does not correspond to the distribution of any particular receptor protein such as calmodulin-stimulated phosphodiesterase and myosin light chain kinase. Furthermore, this ubiquitous distribution of calmodulin distinguishes it from other Ca2+-binding proteins. These findings indicate calmodulin may serve in a general regulatory role in a wide variety of animal species and tissues. The properties of vertebrate calmodulin have been studied in considerable detail. It is an acidic 148-residue protein that contains no tryptophan, cysteine, phosphate, or carbohydrate (25). Based upon a comparison of calmodulin’s amino acid sequence to other Ca2+-binding proteins such as parvalbumin and troponin C, four Ca2 -binding domains have been postulated as well as putative Ca*+-binding amino acid residues within those domains. These Ca2-t-binding domains are separated by regions of a-helical structure. There is homology among the four Ca2 -binding domains in calmodulin, with greater homology between domains 1 and 3 than between 2 and 4. Sequence data also indicate that the primary structure of calmodulin is highly conserved among widely divergent animal and plant species. This apparent structural constraint may be related to the consequence of calmodulin regulating many intracellular functions. Calmodulin binds 4 Ca2+ per mol of calmodulin with high affinity, and studies on the Ca2 -binding properties of calmodulin indicate multiple classes of sites of negative cooperativity (26-30). However, positive cooperativity at low Ca2+ concentrations (31) or equivalent sites (32) has also been described. An extensive study of the binding of cations to calmodulin under a variety of conditions has been performed by Haiech et al. (33).Results from various studies have yielded four different macroscopic binding constants for the different Ca2 binding sites (Table 1). However, Potter et al. (34) believe that calmodulin binds 4 mol Ca2 per mol of calmodulin at essentially the same affinity for each of the four sites. These authors discuss the technical and theoretical limitations in determining Ca2 -binding affinities for a protein that has multiple Ca2 -binding sites. Upon binding Ca2 , calmodulin undergoes conformational changes as indicated by circular dichroism, optical rotary dispersion, nuclear magnetic resonance, and uv-difference spectroscopy (21, 35). Low- and high-affinity Ca2 +

+

+

+

+

+

+

+

+

117

4. CALMODULIN-DEPENDENT PROTEIN KINASES TABLE 1 MACROSCOPIC C A ~ DISSOCIATION CONSTANTS FOR CALMODULIN +

2.0 5.3 5.0 1.5

1.9 3.6 4.5 2.7

7.3 22 17 31

61 83 50 31

(33) (128) (130) (131)

binding sites of calmodulin have been detected by these procedures. Based upon differences in binding properties of the different sites for Ca2+, it has been suggested that various calmodulin-related enzymes might be activated by distinct conformers of calmodulin containing different amounts of Ca2+ (21, 35).

2 . Calmodulin-Dependent Enzymes Cheung (36) and Kakiuchi et al. (37) first reported an activator for the Ca2 stimulated cyclic nucleotide phosphodiesterase. Teo and Wang (26) showed that the activator protein (calmodulin) bound Ca2 ; Teshima and Kakiuchi (38) demonstrated that calmodulin formed a Ca2 -dependent complex with phosphodiesterase. From these and other investigations a general mechanism for the Ca2 -dependent activation of various enzymes by calmodulin has been described (21-24). The two reactions involving the activation process are shown in Eqs. (1) and (2). +

+

+

+

+ CaM Ca,2+CaM* + E nCa2

+

Caz -CaM*

(1)

Caz+CaM*.E*

(2)

+

Calmodulin (CaM) binds Ca2+, which is accompanied by specific conformational changes. The active conformer of calmodulin (CaM*) binds to a receptor enzyme (E) which may be inactive or partially active. The binding of calmodulin to the enzyme results in a conformational change associated with activation or stimulation of catalysis (E*). This general scheme ignores some of the important questions regarding stoichiometries of Ca2 -binding to calmodulin that may be required for activation of a particular enzyme. Calmodulin has four Ca2+ -binding domains and it has been suggested that different classes of receptor proteins may interact with calmodulin at different extents of Ca2+ occupancy in the four different domains (21). It should also be noted that this general scheme is not meant to imply a particular stoichiometry for calmodulin binding to receptor proteins. These particular issues are discussed in detail in the following sections regarding the different protein kinases. +

118


The ability of calmodulin to increase enzyme activities in a Ca2+-dependent manner has been described for many enzymes from a variety of tissues and animal species including a cyclic nucleotide phosphodiesterase (36, 39), brain adenylate cyclase (40), phosphorylase kinase (41), NAD kinases (42, 43), guanylate cyclase (44), membrane ATPases (45-48), a phospholipase activity (49), myosin light chain kinases (50-52), and other Ca2 -dependent protein kinases (53-55). Considering the ubiquitous nature of calmodulin and its ability to affect activation of numerous enzymes, it seems likely that the list of biological processes regulated by calmodulin will continue to grow. Many of these enzymes have not been purified to homogeneity so that quantitative information regarding mechanisms of enzyme activation are limited. As pointed out by others (23), calmodulin stimulation of enzyme activity may not necessarily reflect a physiological function of calmodulin. For example, calmodulin can substitute for another Ca2 -binding protein that regulates enzyme activity (i.e., troponin C regulation of actomyosin ATPase activity) (56, 57). +

+

II. Myosin Light Chain Kinases A.

INTRODUCTION

Eukaryotic cells contain actin, myosin, and related proteins that are of primary importance in muscle contractility and motility of nonmuscle cells (58-64). In general, muscle contraction is thought to consist of sliding of interdigitating thick myosin filaments and thin actin filaments past each other. The driving force for this process is the cyclic attachment and detachment of the globular head region of myosin to the actin filament with the energy for this process provided by ATP (see Volume XVIII, Chapter 13). The release of Ca2+ into the cytoplasm is the primary event in excitationcontraction coupling, and the subsequent binding of Ca2+ to particular sites on regulatory proteins associated with or acting on the contractile apparatus ultimately results in contraction. Although there may be similarities in the properties of interaction between the contractile proteins actin and myosin from different kinds of muscle and nonmuscle cells, the biochemical mechanisms by which Ca2 triggers activation of the contractile elements may be markedly different (63, 65). In nonmuscle cells there may be additional mechanisms for regulating cell motility (58, 59, 66). Phosphorylation and dephosphorylation of myofibrillar proteins has been considered a potentially important biochemical mechanism for regulating contraction since the discovery that myosin (67) and troponin (68, 69) purified from rabbit skeletal muscle were phosphoproteins, and that purified subunits of these proteins could be phosphorylated by different types of protein kinases (70). +

4.

CALMODULIN-DEPENDENT PROTEIN KINASES

119

Myosins purified from all types of vertebrate tissues contain a single class of light chains that have similar biochemical properties. The masses of these myosin light chains range from 18 to 20 kDa, and they are capable of binding divalent cations. All of the light chain subunits from vertebrate tissues that fall into this class are also capable of being phosphorylated by myosin light chain kinase. These myosin light chains from vertebrate tissues have also been referred to as phosphorylatable or P-light chains (71). Early investigations demonstrated that P-light chain of myosin from white, fast-twitch skeletal (67), cardiac (71), red, slow-twitch skeletal (71), and smooth (72) muscles was reversibly phosphorylated and dephosphorylated. Myosin isolated from smooth muscles as well as from nonmuscle cells requires phosphorylation of P-light chain for actin activation of the Mg2 -dependent myosin ATPase activity (73, 74). A more detailed description of the function of the phosphorylation of myosin in regulating actin and myosin interactions may be found in Volume XVIII, Chapter 13. The original investigations by Perry and his colleagues demonstrated a protein kinase present in rabbit skeletal muscle that catalyzed the phosphorylation of Plight chain. This enzyme had biochemical properties that distinguished it from other types of protein kinases such as cyclic nucleotide-dependent protein kinases and phosphorylase kinase, and it was named myosin light chain kinase (75). The enzyme was highly specific for catalyzing the phosphorylation of Plight chains as compared to other protein substrates. It was also dependent upon Ca2 for activity. It was subsequently demonstrated that myosin light chain kinase requires two protein components for activity. Dabrowska and Hartshorne (50) showed that calmodulin was required for the activation of the catalytic subunit of smooth-muscle myosin light chain kinase. At the same time, Yagi et al. (76) found that skeletal-muscle myosin light chain kinase also required calmodulin for activation. Because myosin light chain kinases were dependent upon Ca2 and calmodulin for activity, the possibility was naturally considered that there may be a link between activation of this kinase and muscle contraction. +

+

+

B.

PHYSICOCHEMICAL PROPERTIES

Myosin light chain kinases have been purified to homogeneity from a number of vertebrate muscles. In addition, myosin light chain kinases have been identified and partially purified from several vertebrate nonmuscle tissues. One of the primary problems in establishing the physicochemical properties of the native forms of myosin light chain kinases has been the marked sensitivity of these kinases to limited proteolysis during purification. Daniel and Adelstein (77) found myosin light chain kinase purified from platelets was a 78-kDa enzyme that was not dependent upon calmodulin for activity. Dabrowska and Hartshorne (50) partially purified calmodulin-dependent myosin light chain kinases from both human platelet and bovine brain that were similar sizes, 105 kDa. Hathaway

120

J. T. STULL, M. H . NUNNALLY, AND C. H. MICHNOFF

and Adelstein (51) subsequently reported purification of a human platelet myosin light chain kinase, with an apparent mass of 105 kDa. However, later work by Hathaway et al. (78) suggested that the bovine brain myosin light chain kinase was a 130-kDa species and, presumably, isolation of the smaller brain and platelet kinases was due to proteolysis during the enzyme purification. An early report on the purification of the chicken gizzard smooth-muscle kinase indicated that this enzyme was 105 kDa (79). However, inclusion of protease inhibitors results in purification of larger myosin light chain kinases from gizzard smooth muscle [turkey gizzard, 130 kDa (80, 81); chicken gizzard, 130-136 kDa (82, 83)]. The molecular weight of the kinase determined by sedimentation equilibrium experiments is 124,000, indicating that the enzyme probably exists as a monomer in its native state (80). The turkey gizzard kinase is asymmetrical with a Stokes radius of 75 A, a sedimentation coefficient of 4.45 S, and a frictional ratio of 1.85. It also binds 1 mol calmodulin per mol kinase. Adachi et al. (84) claimed that myosin light chain kinase from chicken gizzard is 136 kDa. This form of the kinase was identified by immunoblots with a monoclonal antibody to the 130-kDa form. Subsequently, Ngai et al. (83) purified this 136-kDa form of the chicken gizzard myosin light chain kinase. The authors proposed that after extraction from the tissue the kinase undergoes proteolytic cleavage resulting in the size reduction from 136 to 130 kDa. They suggest that it is the partially degraded 130-kDa form of the gizzard kinase that has been routinely purified (80, 81). However, the small change in size prevents a definitive conclusion regarding the form of the kinase purified in different laboratories. Myosin light chain kinases have also been purified from mammalian smooth muscle. The mammalian smooth-muscle kinases appear to be as large or larger than the purified chicken and turkey smooth-muscle myosin light chain kinases [steer stomach, 155 kDa (85);steer trachea, 150-160 kDa (86, 87); steer aorta, 142 kDa (88); hog myometrium, 130 kDa (89); hog carotid artery, 140 kDa (J. T. Stull and S. M. Moreland, unpublished observation)]. The reasons for these slight differences in reported mass for the mammalian myosin light chain kinases is not clear, but may reflect animal species differences (90) in addition to variations in determining sizes among different laboratories. However, it seems clear that mammalian myosin light chain kinases, in general, are of slightly larger size than the gizzard kinase. Whether this difference is due to a difference in primary structure or an undetermined posttranslational modification has yet to be determined. There are no published reports on the hydrodynamic properties of mammalian smooth-muscle myosin light chain kinases. The myosin light chain kinases that have been purified from mammalian striated muscle are considerably smaller than the smooth-muscle kinases. The relative masses determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate for rabbit skeletal-muscle myosin light chain kinase


121

range from 77 to 94 kDa (75, 90-96). The native molecular weight of rabbit skeletal-muscle kinase determined by gel filtration and sedimentation equilibrium is between 70,000 and 80,000 (75, 91, 93, 95). One report presented a value of 103,000 for the native molecular weight determined by sedimentation equilibrium (96). Thus, rabbit skeletal-muscle myosin light chain kinase, similar to the gizzard smooth-muscle enzyme, is a monomeric enzyme. In addition, the skeletal-musclekinase is a highly asymmetrical molecule with a Stokes radius of 54 A, sedimentation coefficient of 3.2 S and a frictional ratio of -2.0 (93, 95). Myosin light chain kinase has also been purified from steer cardiac muscle and is reported to be 85 (97) and 94 kDa (98). A Stokes radius of 44 A (97) suggests that the cardiac kinase also is asymmetrical. It should be noted that the reported sizes of the cardiac enzyme are similar to values for the enzyme from rabbit skeletal muscle, which are smaller than the smooth-muscle kinases. Sellers and Harvey (99) reported the purification of myosin light chain kinase from Limulus striated muscle. Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate demonstrated that the kinase was composed of a protein doublet of 39 and 37 I 60

460: 560

(p/p‘)

51; 60 doublet 52 doublet; 63 doublet 51; 60 50 (a); 60 (P) 49 55 50 (a): 60/58 (p/p’) 50 (a); 60/58 (PIP‘)

> 63

600

51 > 6 0

560

52

165: 545

540; 640 540 550 la;4P/P‘

615

The / indicates mass of subunits where the smaller subunit may be due to proteolysis or other modifications. The > indicates which subunit of the designated mass is more abundant.

Reference


145

subunit self-associates into an oligomeric kinase complex that has a mass of approximately 540 to 640 kDa. Bennett ef al. (189) point out that an acid precipitation step (pH 6.1) used by Fukunaga et al. (195) causes precipitation of the 60-kDa (p) subunit of the brain calmodulin-dependent kinase. However, Yamauchi and Fujisawa (199) did not use an acid precipitation step and, therefore, the absence of a 60-kDa subunit cannot be readily explained. The purification of an active kinase that lacks the 60-kDa subunit raises the question of the role of this subunit in relation to kinase activity. The antibody crossreactivity of the 50- and 60-kDa subunits with a monoclonal to the protein kinase provides evidence that the 60-kDa polypeptide is a subunit of the kinase (201). Bennett et al. (189) found that both the 50- and 60-kDa subunits coprecipitated with a monoclonal antibody to the partially purified kinase. Supporting evidence also comes from the partial peptide maps which show both subunits to be structurally related (187, 196, 198). The oligomeric structure has not been clarified in terms of whether the kinase exists in multiple heteropolymers of 50and 60-kDa subunits or whether homopolymers exist in tissues where both subunits are found. There have been no studies involving cross-linking or nondenaturing polyacrylamide gels of the holoenzymes whereby the subunit compositions could be analyzed. The multifunctional calmodulin-dependent protein kinases purified from brain exist as both soluble and particulate enzyme complexes. It appears that the 50and 60-kDa subunits are similar, if not identical, regardless of subcellular localization (Table VI). Kennedy et al. (Z87)extracted the kinase from the particulate membrane fraction by reducing the ionic strength (keeping the solution isoosmotic) with recovery of 70% of the kinase activity. The kinases purified from the soluble and particulate fractions had identical specificities for substrates, effect of pH on kinase activity, and Ca2+ concentrations required for catalysis. McGuinness et al. (201)purified both particulate and soluble calmodulin-dependent protein kinases from rat forebrain and cerebellum. The forebrain enzyme was more highly concentrated in the soluble fraction (50% of total activity), and another 24-45% was extracted from the particulate fraction. In contrast, 20% of the total cerebellar kinase was present in the soluble fraction, and only an additional 10-15% of the activity could be extracted from the particulate fraction. However, the purified kinases were similar in biochemical properties except in the relative ratios of 50- to 60-kDa subunits (Table VI). Several laboratories have identified the major postsynaptic density protein from rat brain as the 50-kDa subunit of the multifunctional calmodulin-dependent protein kinase (188, 202, 203). This membrane-bound protein has been determined to be similar to the 50-kDa subunit of the kinase on the basis of (a) [ 1251]calmodulinbinding; (b) peptide maps of 32P-labeled and '251-labeled 50kDa proteins; and (c) antibody crossreactivity. Goldenring et al. (202) suggested on the basis of two-dimensional tryptic peptide maps, that the membrane-bound

146

I. T. STULL, M. H. NUNNALLY, AND C. H. MICHNOFF

50- and 60-kDa subunits of the major postsynaptic density protein may have slightly different primary structure than the soluble 50- and 60-kDa kinase subunits. A calmodulin-dependent protein kinase isolated from rat brain neuronal nuclear matrix has 50- and 60-kDa autophosphorylated subunits which appear to

be similar to the polypeptides of the postsynaptic densities (204). However, onedimensional phosphopeptide maps and two-dimensional tryptic maps demonstrated that the membrane-bound 50-kDa major postsynaptic density protein was very similar, but not identical to the 50-kDa subunit of the calmodulin-dependent protein kinase from the nuclear matrix. A calmodulin-dependent protein kinase has been purified from a postsynaptosomal cytoskeletal preparation from rat brain that is distinct from the postsynaptic density preparation (200). This cytoskeletal preparation is enriched 2- to 3fold with kinase compared to the typical postsynaptic density preparation. In contrast to the particulate enzymes purified by Kennedy et al. (183, the kinase associated with the cytoskeleton required 8 M urea for extraction. The purified kinase has a 50-kDa subunit comprising most of the protein with a minor 60-kDa subunit. Both subunits bound [‘251]calmodulinand were autophosphorylated. The only apparent difference between this kinase and the more easily extractable kinases was the different relative ratio of subunits, 6 a : 1p (Table VI). It will be essential to determine the primary structure of these calmodulin-dependent kinases to establish whether there are significant differences in primary structure that are important for subcellular distribution. There are specific forms of the multifunctional calmodulin-dependentprotein kinase associated with particular brain regions (201). The a and p subunits appear to be identical in the different forms, but the relative ratio of a:p/p’ subunits are distinct for the forebrain (3a: lp/p’) and cerebellum (la:4p’), respectively (Table VI). These area-specific ratios exist shortly after tissue dissection and do not seem to arise as a result of the purification procedure. The purified kinases have different native masses, Stokes radii, and sedimentation coefficients, indicative of the relative ratios of a : p subunits. There were no differences in the ratio of a to p on either the ascending or descending sides of the protein peaks after gel filtration, thus arguing against homopolymers consisting of a or p subunits, respectively. In summary, there may be subcellular and region-specific forms of the brain multifunctional calmodulin-dependent protein kinase that exist with different ratios of protein subunits. The biological significanceof these differences may be related to different binding affinities for membranes, the cytoskeleton, or kinase substrates.

2 . Other Tissues The similarities in subunit composition and substrate specificity between the brain multifunctional calmodulin-dependentprotein kinase and a number of other

147

4. CALMODULIN-DEPENDENT PROTEIN KINASES TABLE VII STRUCTURAL PROPERTIES

Tissue and animal species Brain, rat

Heart, steer( Skeletal muscle, rabbit Liver, rabbit Ratd Pancreas, rate Erythrocyte, turkey' Invertebrate neuronal,

OF

MULTIFUNCTIONAL CALMODULIN-DEPENDENT PROTEIN KINASES FROM DIFFERENT TISSUES

Subunit massa (kDa)

Relative compositionb

Native mass (kDa)

Reference

50-55 60158 55-51 13-15 54 58/59 5 1 I53 50153 56/57 51 50/54 58 50-5 1

Variable

165-650

(See Table VI)

55 > 73

100;900

( I92,207)

4(58);l(54)

696 215; 500 300

58

300;600

> 50

Aplysiae

Electric organ, TorpedoC

54 62

The / indicates mass of subunit where the smaller subunit may be due to proteolysis or other modifications. The > indicates which subunit of the designated mass is more abundant. Not homogeneous. Comigrated with rabbit enzyme. Identified by monoclonal antibodies.

calmodulin-dependent protein kinases suggest that all of these enzymes may represent a class of homologous, multifunctional calmodulin-dependent protein kinases. In fact, several groups of investigators report antibody crossreactivity between the rat brain multifunctional calmodulin-dependent protein kinase and kinases isolated from other animal species and tissues. Antibodies to the rat brain enzyme have been shown to cross-react with the kinase from rabbit skeletal muscle (205,206) and Aplysia californica (194). Very similar peptide maps were observed with the Aplysia 50-kDa polypeptide compared to the 50-kDa subunit from rat brain kinase (194). In addition, the kinases identified in other tissues bind ['251]calmodulin and are autophosphorylated. Table VII summarizes some of the structural properties of the multifunctional calmodulin-dependent protein kinase from a number of tissues and animal species. In general, multifunctional calmodulin-dependentprotein kinases from a wide variety of animal species (mammalian, avian, and invertebrate) are composed of either two subunits of approximately 50-55 and 60-75 kDa or the single lowermolecular-mass subunit, 50-55 kDa. Again it is necessary to consider whether

148

J . T. STULL, M. H. NUNNALLY, AND C. H. MICHNOFF

differences in the purification procedures could account for the variable subunit compositions. In the case of kinases from rabbit and rat liver, rat pancreas, and Aplysia neuronal tissue, only the rabbit liver kinase purified by Ahmad et al. (183) was subjected to an acidic pH treatment during purification. As discussed previously, lowering the pH to 6.1 may cause precipitation of the larger subunit (189). The rabbit and rat liver kinases purified by Payne et al. (184) and Schworer er al. (185) were also reported to have a single subunit, and no acidic pH treatment was used. Therefore, the relative ratio of the two subunits in multifunctional calmodulin-dependent protein kinases appears to be dependent upon the type of tissue as well as animal species. The multifunctional calmodulin-dependent protein kinases purified from nonneuronal tissue are present predominantly in the soluble cell fraction (183, 184, 186, 192, 193, 207). A significant portion of the kinase activity in Torpedo electric organ (190) and neuronal tissue of Aplysia ( 194) appears in the particulate fraction as well as in the soluble fraction, similar to the distribution of the enzyme in rat brain. The variability in reported sizes of the smaller and larger subunit of nonneuronal multifunctional calmodulin-dependent protein kinases may be due simply to differences in relative electrophoretic mobility under different buffer systems. For example, multifunctional calmodulin-dependent protein kinase purified from rat liver was reported to have a molecular mass of 56-57 kDa (185) and, therefore, appears to be different from the rabbit liver kinase, 51-53 kDa. However, these authors reported that the two forms from rat and rabbit liver comigrated during polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate using the discontinuous Laemmli (208) buffer system. Similarly, Woodgett et al. (206) reported that the multifunctional calmodulin-dependent protein kinase from rabbit skeletal muscle comigrates with the rat brain kinase, but that the apparent mass of the major component for both is either 50 or 55 kDa, depending upon the electrophoretic buffer system. Therefore, these minor differences in reported masses may reflect variable electrophoretic mobilities of the subunits under different conditions. The differences in reported native masses may also be due to variable experimental conditions. For example, different ionic strengths used during the determination of the Stokes radius or sedimentation coefficient may have significant effects. Kloepper and Landt (207) reported two different native masses for the multifunctional kinase purified from bovine heart, either 100 or 900 kDa, as determined by sucrose-density gradient centrifugation. The relative proportion of total activity in either peak, as well as enzyme stability, was dependent upon the NaCl concentration. However, because this kinase preparation was only partially pure, it was not possible to rule out interactions with other proteins that may cause the great size differences. In contrast, Woodgett et al. (286) found that the Stokes radius determined by gel filtration of the rabbit skeletal-muscle kinase

4. CALMODULIN-DEPENDENT PROTEIN KlNASES

149

was unaffected by ionic strength up to 0.5 M NaCI. The native enzyme was shown by transmission electron microscopy to be a 10 nm diameter hexameric ring (186). Because the estimated mass was 700 kDa, the authors concluded that the structure was a dodecamer, composed of two hexameric rings stacked one upon the other. The suggestion was made that the 300-kDa form present in other tissues may be composed of a single hexameric ring. The enzyme from rabbit liver had an apparent native mass of either 500 or 250 kDa as determined by gel filtration or sucrose-density gradient centrifugation under conditions of identical ionic strength (183). Similar results were obtained with multifunctional calmodulin-dependent protein kinase from rat pancreas (209). Further characterization of the hydrodynamic properties of multifunctional calmodulin-dependent protein kinases needs to be performed to clarify the native structures. C. C A ~ AND + CALMODULIN ACTIVATION The multifunctional protein kinases are dependent upon both Ca2+ and calmodulin for activity with a variety of protein substrates (35, 182), but much less is known about the specific, biochemical properties of activation by calmodulin as compared to similar information for myosin light chain kinases. The enzyme activity is inhibited in a competitive fashion by various calmodulin antagonists (55, 194). Measurement of ['251]calmodulinbinding via the gel overlay method shows that binding polypeptides comigrate with both 50- and 60-kDa protein subunits in purified kinase preparations and binding is dependent upon the presence of Ca2+ (196, 200, 201). These experimental approaches demonstrate qualitatively the Ca2 .calmodulin dependence of the multifunctional calmodulin-dependent protein kinase activity and identify calmodulin binding subunits. The concentration of calmodulin required for half-maximal activation is, in general, greater than the concentrations of calmodulin required for half-maximal activation of myosin light chain kinases (Table VIII). Reported values determined under a variety of experimental conditions vary from 0.012 to 0.4 pM (Table VIII). Some investigators have also measured the concentration of Ca2+ required for half-maximal activation of the multifunctional calmodulin-dependent protein kinases. These experiments have been performed at a particular calmodulin concentration and Ca2+ values range from 0.8 to 4.0 pA4 (Table VIII). Thus, the concentrations of Ca2+ and calmodulin required for halfmaximal activation of the multifunctional calmodulin-dependent protein kinases are within a range that is biologically relevant. However, additional studies need to be performed with the purified multifunctional calmodulin-dependent protein kinases to determine the stoichiometry of Ca: *calmodulinbinding to the holoenzyme required for activation and how many divalent cation binding sites in calmodulin need to be occupied by Ca2+ for enzyme activation. It will be interesting to compare the energy coupling values for Ca2+, calmodulin, sub+

+

150

J. T. STULL, M. H. NUNNALLY, AND C. H. MICHNOFF TABLE VIIl C A+~ AND CALMODULIN ACTIVATION OF MULTIFUNCTIONAL CALMODULIN-DEPENDENT PROTEIN KINASES Ko,5Calmodulin at

[Ca2+Ifrce

(W)

(W)

(W)

(W)

Reference

4.0 0.8 I .9

0.6 6.0 0.3 0.1

300 5 200 500 I20 400

(187)

1.6

0.40 0. I6 0.012 0.10 0.010 0.080

K&aZ+

at [Calmodulin]

(257) (195) (54)

(199) (183)

strates, and multifunctional calmodulin-dependent protein kinase with the reported values for myosin light chain kinase from rabbit skeletal muscle.

D. CATALYTIC PROPERTIES The multifunctional calmodulin-dependent protein kinase is present in a number of different tissues as discussed previously. Of the tissues examined, the greatest kinase activities are found in brain, and possibly constitute 0.1-0.4% of total brain protein (187, 189). Regional variations of calmodulin-dependentprotein kinase activity are found in brain (210-212) and the enzyme comprises approximately 0.4 and 0.2% of total protein in rat forebrain and cerebellum regions, respectively (201).In brain and nonneuronal tissue, the relative catalytic activities are brain (loo%), spleen (25%), heart (12%), adrenal (9%), skeletal muscle (4%), and liver and kidney ( 2 > lb, respectively (240). As discussed previously, specificity of protein kinases for protein substrates is significantly influenced by basic residues in the primary structure around the phosphorylated site (156, 159, 160, 227). Amino acid sequences of sites phosphorylated by the multifunctional calmodulin-dependent protein kinase are shown in Table XIII. All phosphorylation sites shown in Table XI11 have one or more basic amino acids located two or three residues toward the amino terminus from the phosphorylatable serine; this property is similar to sites phosphorylated by CAMP-dependent protein kinase. Synthetic peptides of skeletal-muscle glycogen synthase or smooth-muscle myosin P-light chains were used as sub-

TABLE XI11 AMINOACIDSEQUENCES SURROUNDING SITES PHOSPHORYLATED BY MULTIFUNCTIONAL CALMODULIN-DEPENDENT PROTEINKINASES~ Source

Sequence

Reference

Skeletal-muscle glycogen synthase P L S R T L V S S L site 2 Skeletal-muscle glycogen synthase S G G S K R S N V D T S site Ib Smooth-muscle P-light chain K K R P Q R A T S N V F S S R K L

Phenylalanine hydroxylase Liver pyruvate kinase

S

For amino acid abbreviations. see Table V.

(260.26/)

(252)

B F G Z Z (262).

L R R A S L ~

(258,259)

(263)


157

strates for multifunctional calmodulin-dependent protein kinase (241). The apparent K, value for a synthetic peptide representing glycogen synthase phosphorylation site 2 in Table XI11 was 2-fold greater than the K, value for native glycogen synthase (241). However, the V,,, value for the synthetic peptide was 10-fold greater than the native protein substrate. A synthetic peptide identical with smooth-muscle myosin P-light chain was phosphorylated with a lower K , value and a higher V,,, value than values obtained with native gizzard-muscle myosin P-light chain. The arginine, three residues toward the amino terminus from the phosphorylatable serine in glycogen synthase and myosin P-light chain peptides, is an important determinant for phosphorylation by multifunctional calmodulin-dependent protein kinase.

E. PHOSPHORYLATION The protein subunits of multifunctional calmodulin-dependent protein kinases are phosphorylated in the presence of Ca2+ and calmodulin (183, 186, 189, 195, 198, 204). Multiple sites in each protein subunit appear to be phosphorylated. Bennett et al. (189) reported the incorporation of 2 mol phosphate per mol of 50kDa subunit and 3 mol phosphate per mol 60-kDa subunit with brain kinases. Approximately 4 mol phosphate are incorporated into the subunits of liver calmodulin-dependent protein kinase (183). Similarly, both protein subunits of skeletal-muscle calmodulin-dependent protein kinase are autophosphorylated with an estimated 4-5 mol phosphate per mol protein (186). Serine is the major amino acid autophosphorylated (196). Autophosphorylation of multifunctional calmodulin-dependent protein kinases decreases subunit mobility in sodium dodecyl sulfate polyacrylamide gels (189, 198, 201, 204). After phosphorylation in vitro, the 50- and 60-kDa subunits change to 53-54 and 64-68 kDa, respectively (189, 198, 201, 204). Similar effects of phosphorylation on protein mobility in sodium dodecyl sulfate polyacrylamide gels are observed with type I1 regulatory subunit of CAMP-dependent protein kinase (242-244) and glycogen synthase kinase 3 (245). Thus, the reported differences in protein subunit mobility of multifunctional calmodulindependent protein kinases could be due, in part, to different amounts of phosphate incorporated into the polypeptides. However, there have been no chemical measurements of protein-bound phosphate in purified nonphosphorylated multifunctional calmodulin-dependent protein kinases. Attempts have been made to identify the catalytic subunit of multifunctional calmodulin-dependent protein kinase. As a result of autophosphorylation studies and binding studies with 8-N3-ATP (198, 246), both subunits were concluded to contain active sites. These studies cannot disregard, however, possible nonspecific interactions that may occur between the nitrene in 8-N3-ATP and the protein (247). Furthermore, the data do not discount the possibility of a single

158

J. T. STULL, M. H. NUNNALLY, AND C. H . MICHNOFF

type of catalytic subunit which autophosphorylates and phosphorylates another type of subunit. The role of autophosphorylation on the biochemical properties of multifunctional calmodulin-dependentprotein kinases is not clearly defined. Shields et al. (246) reported 50-70% increased binding of [ 1251]calmodulin,measured by the gel overlay method, to both the 50- and 60-kDa polypeptides after autophosphorylation in synaptic junction fractions. LeVine et al. (248) examined calmodulin binding to a neuronal cytoskeleton preparation enriched with the kinase. Similar to the observations of Shields et al. (246), [1251]calmodulin binding was greater to phosphorylated calmodulin-dependentprotein kinase from enriched cytoskeleton by the gel overlay method (248).However, [ 1251]calmodulin binding affinity to phosphorylated calmodulin-dependent protein kinase in enriched cytoskeleton decreased 2-fold when measurements were performed in a buffered solution (248). Results from these studies show opposite effects of calmodulin-dependent protein kinase autophosphorylation on calmodulin-binding properties. These differences may be a result of calmodulin binding measured with denatured or native calmodulin-dependent protein kinase by gel overlay or in solution, respectively. Additional studies are required for a thorough characterization of calmodulin-binding properties of autophosphorylated calmodulin-dependent protein kinase, as well as an evaluation of the biological significance.

F. BIOLOGICAL SIGNIFICANCE OF MULTIFUNCTIONAL CALMODULIN-DEPENDENT PROTEINKINASES There have been many reviews written about the roles Ca2+- and CAMPdependent protein phosphorylation play in the regulation of neuronal function (182, 249-252). For example, changes in the intracellular concentrations of Ca2+ and cyclic nucleotides have been implicated in the regulation of neurotransmitter biosynthesis and release, RNA transcription and protein translation, synaptic vesicle fusion, cytoskeletal protein organization, and ion channels, and it has been proposed that most, if not all, of these processes are regulated by CAMPor Ca2 -dependent protein kinases. The reader should refer to the many chapters on control of specific enzymes and biological processes in this volume for more details. Ouimet e f al. (212) demonstrated by immunocytochemistry that the brain multifunctional calmodulin-dependent protein kinase is associated in situ with subcellular structures that contain known in vifro substrates, such as synapsin I and the microtubule-associated protein, MAP-2. Biochemical measurements suggest that calmodulin-dependentprotein kinase activity is associated with isolated synaptosomes, neuronal nuclei and a cytoskeletal protein fraction. Clearly, the multifunctional calmodulin-dependent protein kinase is located within the brain at sites accessible to protein substrates identified in vitro. +


159

Calcium regulation of biological processes is not limited to neuronal tissue. Neural stimulation of muscle and hormonal stimulation of liver and other tissues are examples of external signals that may be mediated intracellularly by the second messenger, Ca2 . The multifunctional calmodulin-dependent protein kinase has been found in a large number of vertebrate tissues. The large number of protein substrates phosphorylated by this kinase suggests that it is a general protein kinase that mediates Ca2 -regulated processes in many cell types. However, the specific role that this kinase plays in any particular cell depends on a number of factors including subcellular localization of the kinase and substrate availability. It will be necessary to satisfy the criteria of Krebs and Beavo (146) to establish the biological importance of phosphorylation for each protein phosphorylated by a multifunctional calmodulin-dependent protein kinase. +

+

ACKNOWLEDGMENTS The authors wish to express their sincere appreciation to many colleagues who sent reprints and preprints for this chapter. In addition, the authors acknowledge the dedicated, superb assistance of Nancy Bryant and Kathy Perdue in preparing this manuscript. Support has been generously provided from the National Institutes of Health (HL23990, HL26043, HL06296) and the Muscular Dystrophy Association for research described in this chapter.

REFERENCES Ringer, S. (1883). Pructitioner31, 81. Ringer, S. (1883). J. Physiol. (London) 4, 29. Ringer, S. (1883). J. Physiol. (London) 4, 222. Locke, F. S. (1894). Zentralbl. Physiol. 8, 166. 5 . Campbell, A. K. (1983). “Intracellular Calcium-Its Universal Role as Regulator.” Wiley, New York. 6. Kretsinger, R. H. (1983). In “Calcium and Cell Function” (W. Y. Cheung, ed.), Vol. 4, p. 212. Academic Press, New York. 7. Bailey, K. (1942). BJ 36, 121. 8. Needham, D. M. (1942). BJ 36, 1 13. 9. Heilbrunn, L. V., and Wiercinski, F. J. (1947). J. Cell. Comp. Physiol. 29, 15. 10. Ebashi, S. (1963). Nurure (London) 200, 1010. 11. Ebashi, S., Kodama, A., and Ebashi, F. (1968). J. Biochem. (Tokyo) 64, 465. 12. Meyer, W. L., Fischer, E. H., and Krebs, E. G. (1964). Biochemistry 3, 1033. 13. Ozawa, E., Hosoi, K., and Ebashi, S. (1967). J . Biochern. (Tokyo) 61, 531. 14. Sutherland, E. W., Robinson, G. S., and Butcher, R. W. (1968). Circulation 37, 279. 15. Rasmussen, H. (1970). Science 170, 404. 16. Rasmussen, H. (1983). I n “Calcium and Cell Function” (W. Y.Cheung, ed.), Vol. 4, p. 1. Academic Press, New York. 17. Rasmussen, H., and Barrett, P. Q. (1984). Physiol. Rev. 64, 938. 18. Maruta, H., and Korn, E. D. (1977). JBC 252, 8329. 19. Kuczmarski, E. R., and Spudich, J. A. (1980). PNAS 77, 7292. I. 2. 3. 4.

160

J. T. STULL, M. H. NUNNALLY, ANDC. H. MICHNOFF

Hammer, J. A,, 111, Albanesi, J. P., and Kom, E. D. (1983). JBC 258, 10168. Klee, C. B., Crouch, T. H., and Richman, P. G. (1980). Annu. Rev. Biochem. 49, 489. Cheung, W. Y. (1980). Science 207, 19. Van Eldik, L. J., Zendegui, J. G., Marshak, D. R., and Watterson, D. M. (1982). Int. Rev. Cytology 77, 1. 24. Klee, C. B., and Vanaman, T. C. (1982). Adv. Protein Chem. 35, 213. 25. Watterson, D. M., Sharief, F., and Vanaman, T. C. (1980). JBC 255, 962. 26. Teo, T. S., and Wang, I. H. (1973). JBC 248, 5950. 27. Lin, Y. M., Liu, Y. P.,and Cheung, W. Y. (1974). JBC 249, 588. 28. Watterson, D. M., Harrelson, W. G., Jr., Keller, P. M., Sharief, F., and Vanaman, T. C. (1976). JBC 251, 4501. 29. Klee, C. B. (1977). Biochemistry 16, 1017. 30. Wolff, D. J., Pokier, P. G., Brostrom, C. O., and Brostrom, M. A. (1977). JBC 252, 4108. 31. Crouch, T. H., and Klee, C. B. (1980). Biochemistry 19, 3692. 32. Dedman, J. R., Potter, J. D., Jackson, R. L., Johnson, J. D., and Means, A. R. (1977). JBC 252, 8415. 33. Haiech, J., Klee, C. B., and Demaille, J . G. (1981). Biochemistry 20, 3890. 34. Potter, J. D., Strang-Brown, P., Walker, P. L., and lida, S . (1983). “Methods in Enzymology,” Vol. 102, p. 135. 35. Manalan, A. S . , and Klee, C. B. (1984). Adv. Cyclic Nucleotide Protein Phosphorylution Res. 18, 227. 36. Cheung, W. Y. (1970). BBRC 38, 533. 37. Kakiuchi, S., Yamazaki, R., and Nakajima, H. (1970). Proc. Jpn. Acud. 46, 587. 38. Teshima, Y., and Kakiuchi, S. (1974). BBRC 56, 489. 39. Kakiuchi, S., and Yamazaki, R. (1970). BBRC 41, 1104. 40. Brostrom, C. 0.. Huang, Y.-C., Breckenridge, B. McL., and Wolff, D. J. (1975). PNAS 72, 64. 41. Cohen, P., Burchell, A., Foulkes, J. G., Cohen, P. T. W., Vanaman, T. C., and Naim, A. C. (1978). FEES Lett. 92, 287. 42. Epel, D., Patton, C., Wallace, R. W., and Cheung, W. Y. (1981). Cell 23, 543. 43. Anderson, J. M., Charbonneau, H., Jones, H. P., McCann, R. O., and Cormier, M. J. (1980). Biochemistry 19, 3 1 13. 44. Nagao, S., Suzuki, Y., Watanabe, Y., and Nozawa, Y. (1979). BBRC 90, 261. 45. Gopinath, R. M., and Vincenzi, F. F. (1977). BBRC 77, 1203. 46. Jarrett, H. W., and Penniston, J. T. (1977). BBRC 77, 1210. 47. Blum, J. J., Hayes, A., Jamieson, G. A., and Vanaman, T. C. (1980). J. Cell Biol. 87, 386. 48. Tuana, B. S . , and MacLennan, D. H. (1984). JBC 259, 6979. 49. Wong, P. Y.-K., and Cheung, W. Y. (1979). BBRC 90, 473. 50. Dabrowska, R., and Hartshome, D. J. (1978). BBRC 85, 1352. 51. Hathaway, D. R., and Adelstein, R. S. (1979). PNAS 76, 1653. 52. Yazawa, M., Kuwayama. H., and Yagi, K. (1978). J. Biochem (Tokyo) 84, 1253. 53. DeLorenzo, R. J., Freedman, S. D., Yohe, W. B., and Maurer, S. C. (1979). PNAS 76, 1838. 54. Payne, M. E., and Soderling, T. R. (1980). JBC 255, 8054. 55. Kennedy, M. B., and Greengard, P. (1981). PNAS 78, 1293. 56. Amphlett, G. W., Vanaman, T. C . , and Perry, S. V. (1976). FEBS Lett. 72, 163. 57. Dedman, J. R., Potter, J. D., and Means, A. R. (1977). JBC 252, 2437. 58. Kom, E. D. (1978). PNAS 75, 588. 59. Stossel, T. P. (1978). Annu. Rev. Med. 29, 427 60. Taylor, E. W. (1979). CRC Crit. Rev. Biochem. 6, 103. 61. Weingrad, S. (1979). In “Handbook of Physio/ogy” (R. M. Berne, N. Sperelakis, and S. R. 20. 21. 22. 23.


161

Geiger, eds.), 2nd ed., Sect. 2, Vol. I, p. 393. Am. Physiol. Soc., Bethesda, Maryland. Harrington, W. F., and Rodgers, M. E. (1984). Annu. Rev. Biochem. 53, 35. Leavis, P. C., and Gergely, J. (1984). CRC Crit. Rev. Biochem. 16, 235. Eisenberg, E., and Hill, T. L. (1985). Science 227, 999. Kamm, K. E., and Stull, J. T. (1985). Annu. Rev. Pharmacol. Toxicol. 25, 593. Means, A. R., and Dedman, J . R. (1980). Nature (London) 285, 73. Perrie, W. T., Smillie, L. B., and Perry, S. V. (1973). BJ 135, 151. England, P. J., Stull, J. T., and Krebs, E. G . (1972). JBC 247, 5275. Stull, J. T., Brostrom, C. O., and Krebs, E. G. (1972). JBC 247, 5272. Stull, J. T. (1980). Adv. Cyclic Nucleotide Res. 12, 39. Frearson, N., and Perry, S. V. (1975). BJ 151, 99. Frearson, N., Focant, B. W. W., and Perry, S. V. (1976). FEBS Lett. 63, 27. Adelstein, R. S., and Eisenberg, E. (1980). Annu. Rev. Biochem. 49, 921. Cooke, R., and Stull, J . T. (1981). In “Cell and Muscle Motility” (R. M. Dowben and J. W. Shay, eds.), Vol. 1, p. 99. Plenum, New York. 75. Pires, E. M. V., and Perry, S. V. (1977). BJ 167, 137. 76. Yagi, K., Yazawa, M., Kakiuchi, S., Ohshima, M., and Uenishi, K. (1978). JBC 253, 1338. 77. Daniel, J. L., and Adelstein, R. S. (1976). Biochemistry 15, 2370. 78. Hathaway, D. R., Adelstein, R. S., and Klee, C. B. (1981). JBC 256, 8183. 79. Dabrowska, R., Aromatorio, D., Sherry, J. M. F., and Hartshome, D. J. (1977). BBRC 78, 1263. 80. Adelstein, R. S . , and Klee, C. B. (1981). JBC 256, 7501. 81. Walsh, M. P., Hinkins, S., Dabrowska, R., and Hartshorne, D. J. (1983). “Methods in Enzymology,” Vol. 99, p. 279. 82. Uchiwa, H., Kato, T., Onishi, H., Isobe, T., Okuyama, T., and Watanabe, S. (1982). J. Biochem. (Tokyo) 91, 273. 83. Ngai, P. K., Carmthers, C. A,, and Walsh, M. P. (1984). BJ 218, 863. 84. Adachi, K., Carruthers, C. A., and Walsh, M. P. (1983). BBRC 115, 855. 85. Walsh, M. P., Hinkins, S., Flink, 1. L., and Hartshorne, D. J . (1982). Biochemistry 21,6890. 86. Miller, J. R., Silver, P. J., and Stull, J. T. (1983). Mol. Pharmacol. 24, 235. 87. Nishikawa, M., de Lanerolle, P., Lincoln, T. M., and Adelstein, R. S. (1984). JBC 259,8429. 88. Vallet, B., Molla, A,, and Demaille, J. G. (1981). BBA 674, 256. 89. Higashi, K., Fukunaga, K., Matsui, K., Maeyama, M., and Miyamoto, E. (1983). BBA 747, 232. 90. Nunnally. M. H., and Stull, J. T. (1984). JBC 259, 1776. 91. Yazawa, M., and Yagi, K. (1978). J. Biochem. (Tokyo) 84, 1259. 92. Blumenthal, D. K., and Stull, J. T. (1980). Biochemistry 19, 5608. 93. Crouch, T. H., Holroyde, M. J., Collins, J. H., Solaro, R. J., and Potter, J . D. (1981). Biochemistry 20, 63 18. 94. Edelman, A. M., and Krebs, E. G. (1982). FEBS Lett. 138, 293. 95. Mayr, G. W., and Heilmeyer, L. M. G., Jr. (1983). Biochemistry 22, 4316. 96. Nagamoto, H., and Yagi, K. (1984). J. Biochem. (TokyoJ 95, 1119. 97. Walsh, M. P., Vallet, B., Autric, F., and Demaille, J. G. (1979). JBC 254, 12136. 98. Wolf, H., and Hofmann, F. (1980). PNAS 77, 5852. 99. Sellers, J. R., and Harvey, E. V. (1984). Biochemisrry 23, 5821. 100. Walsh, M. P., and Guilleux, J. C. (1981). Adv. Cyclic Nucleoride Res. 14, 375. 101. Guerriero, V., Jr., Rowley, D. R., and Means, A. R. (1981). Cell 27, 449. 102. Nunnally, M. H., Rybicki, S. B., and Stull, J. T. (1985). JBC 260, 1020. 103. Cleveland, D. W., Fischer, S. G . , Kirschner, M. W., and Laemmli, U. K. (1977). JBC 252, 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.

1102.

162


104. Srivastava, S . , and Hartshome, D. J. (1983). EERC 110, 701. 105. Blumenthal, D. K., Takio, K., Edelman, A. M., Charbonneau, H., Titani, K., Walsh, K. A., and Krebs, E. G. (1985). PNAS 82, 3187. 106. Cox, J. A., Comte, M., Fitton, J. E., and DeGrado, W. F. (1985). JBC 260, 2527. 107. Malencik, D. A,, and Anderson, S. R. (1984). Biochemistry 23, 2420. 108. Adelstein, R. S., Conti, M. A., Hathaway, D. R., and Klee, C. B. (1978). JBC 253, 8347. 109. Walsh, M. P., Dabrowska, R.,Hinkins, S., and Hartshome, D. J . (1982). Biochemistry 21, 1919. 110. Foyt, H. L., Guemero, V., Jr., and Means, A. R. (1985). JBC 260, 7765. 111. Scordilis, S. P., and Adelstein, R. S. (1978). JBC 253, 9041. 112. Bremel, R. D., and Shaw, M. E. (1978). FEES Lett. 88, 242. 113. Walsh, M. P., Cavadore, I. C., Vallet, B., and Demaille, J. G.(1980). Can. J . Biochem. 58, 299. 114. Tanaka, T., Naka, M.. and Hidaka, H. (1980). BERC 92, 313. 115. Naim, A. C., and Perry, S. V. (1979). BJ 179, 89. 116. Blumenthal, D. K., and Stull, J . T.(1982). Biochemisrry 21. 2386. 117. Bhalla, R. C., Sharma, R. V., and Gupta, R. C. (1982). BJ 203, 583. 118. Malencik, D. A., Anderson, S. R., Bohnert, J. L., and Shalitin, Y. (1982). Biochemistry 21, 403 1. 119. Johnson, J. D., Holroyde, M. J., Crouch, T. H., Solaro, R. J., and Potter, J . D. (1981). JEC 256, 12194. 120. Nishikori, K., Weisbrodt, N. W., Sherwood, 0. D., and Sanborn, B. M. (1983). JBC 258, 2468. 121. Huang, C. Y., Chau, V., Chock, P. B., Wang, I. H., and Sharma, R. K. (1981). PNAS 78, 871. 122. Stull, J . T., Sanford, C. F., Manning, D. R., Blumenthal, D. K., and High, C. W. (1981).In “Protein Phosphorylation” (0.M. Rosen and E. G.Krebs, eds.), Vol. 8, p. 823. Cold Spring Harbor Press, Cold Spring Harbor, New York. 123. Cox, J. A., Malnoe, A., and Stein, E. A. (1981). JEC 256, 3218. 124. Cox, J. A., Comte, M., Malnoe, A,, Burger, D., and Stein, E. A. (1984). In “Metal Ions in Biological Systems” (H. Sigel, ed.), Vol. 17, p. 215. Dekker, New York. 125. Cox, J . A., Comte, M., and Stein, E. A. (1982). PNAS 79, 4265. 126. Malnoe, A., Cox, J. A., and Stein, E. A. (1982). BBA 714, 84. 127. Burger, D., Stein, E. A , , and Cox, J. A. (1983). JEC 258, 14733. 128. Keller, C. H., Olwin, B. B., LaPorte, D. C., and Storm, D. R. (1982). Biochemistry 21, 156. 129. Olwin, B . B., Edelman, A. M., Krebs, E. G.,and Storm, D. R. (1984). JBC 259, 10949. 130. Cox, J. A. (1984). FP 43, 3000. 131. Chau, V., Huang, C. Y., Chock, P. B., Wang, J. H., and Sharma, R. K. (1982).In “Calmodulin and Intracellular Ca+ Receptors” (S. Kakiuchi, H. Hidaka, and A. R. Means, eds.), p. 199. Plenum, New York. 132. Stull, J. T., Nunnally, M. H., Moore, R. L., and Blumenthal, D. K. (1985). Adv. Enzyme Regul. 23, 123. 133. Moore, R. L., and Stull, J. T. (1984). Am. J . Physiol. 247, C462. 134. LaPorte, D. C., Wierman, B. M., and Storm, D. R. (1980). Biochemistry 19, 3814. 135. Tanaka, T., and Hidaka, H. (1980). JBC 255, 11078. 136. Weiss, B . , and Levin, R. M. (1978). Adv. Cyclic Nucleotide Res. 9, 285. 137. Asano, M., and Hidaka, H. (1984). In “Calcium and Cell Function” (W. Y. Cheung, ed.), Vol. 5 , p. 123. Academic Press, New York. 138. Zimmer, M., Gobel, C., and Hofmann, F. (1984). FEES Lett. 139, 295. 139. Maulet, Y., and Cox, J . A. (1983). Biochemistry 22, 5680. +


163

140. Barnette, M. S., Daly, R., and Weiss, B. (1983). Biochem. Pharmacol. 32, 2929. 141. Newton, D. L., Olderwurtel, M. D., Krinks, M. H., Shiloach, J., and Klee, C. B. (1984).JBC 259, 4419. 142. Ni, W.-C., and Klee, C. B. (1985). JBC 260, 6974. 143. Newton, D. L., and Klee, C. B. (1984). FEBS Lett. 165, 269. 144. Hartshorne, D. J . , and Mrwa, U. (1980). In “Calcium Binding Proteins: Structure and Function” (F. L. Siegel, E. Carafoli, R. H. Kretsinger, D. H. MacLennan, and R. H. Wasserman, eds.), p. 255. ElsevierlNorth-Holland, Amsterdam. 145. Tawata, M., Kobayashi, R., and Field, J. B. (1983). Endocrinology (Baltimore) 112, 701. 146. Krebs, E. G., and Beavo, J. A. (1979). Annu. Rev. Biochem. 48, 923. 147. Persechini, A., and Hartshorne, D. J. (1983). Biochemistry 22, 470. 148. Sellers, J. R., Chock, P. B., and Adelstein, R. S. (1983). JBC 258, 14181. 149. Ikebe, M., Ogihara, S., and Tonomura, Y. (1982). J . Biochem. (Tokyo)91, 1809. 150. Sobieszek, A. (1985). Biochemistry 24, 1266. 151. Scordilis, S. P., Anderson, J. L., Pollack, R., and Adelstein, R. S. (1977). J. Cell Biol. 74, 940. 152. Bourguignon, L. Y. W., Nagpal, M. L., Balazovich, K., Guerriero, V., and Means, A. R. (1982). J . Cell Biol. 95, 193. 153. Singh, T. I., Akatsuka, A,, and Huang, K. P. (1983). FEBS Lett. 159, 217. 154. Persechini, A., and Stull, J. T. (1984). Biochemistry 23, 4144. 155. Kemp, B. E., Pearson, R. B., and House, C. (1982). JBC 257, 13349. 156. Kemp, B. E., Pearson, R. B., and House, C. (1983). PNAS 80, 7471. 157. Kemp, B. E., and Pearson, R. B. (1985). JBC 260, 3355. 158. Michnoff, C. H., Stull, J. T., and Kemp, B. E. (1986). JBC (in press). 159. Chan, K. F. J., Hurst, M. O., and Graves, D. 3. (1982). JBC 257, 3655. 159a. Gauss, A., Mayr, G. W., and Heilmeyer, L. M. G. (1985). EJB 153, 327. 160. Kemp, B. E., Bylund, D. B., Huang, T.-S., and Krebs, E. G. (1975). PNAS 72, 3448. 161. Glass, D. B., and Krebs, E. G. (1982). JBC 257, 1196. 162. Mrwa, U.,and Hartshorne, D. J. (1980). FP 39, 1564. 163. Foyt, H. L., and Means, A. R. (1985). J. CyclicNucleoride Prorein Phosphoryiurion Res. 10, 143. 164. Conti, M. A., and Adelstein, R. S. (1980). FP 39, 1569. 165. Hathaway, D. R., Eaton, C. R., and Adelstein, R. S. (1981). Nature (LondonJ 291, 252. 166. Conti, M. A., and Adelstein, R. S. (1981). JBC 256, 3178. 167. Adelstein, R . S., Sellers, J. R., Conti, M. A,, Pato, M. D., and de Lanerolle, P. (1982). FP 41, 2873. 168. de Lanerolle, P., Nishikawa, M., Yost, D. A., and Adelstein, R. S. (1984). Science 223, 1415. 169. Silver, P. J., and Stull, J. T. (1982). JBC 257, 6145. 170. Gerthoffer, W. T., Trevethick, M. A., and Murphy, R. A. (1984). Circ. Res. 54, 83. 171. Jones, A. W., Bylund, D. B., and Forte, L. R. (1984). Am. J . Physiol. 246, H306. 172. Walsh, M. P., Bridenbaugh, R., Hartshorne, D. J., and Kerrick, W. G . L. (1982). JBC 257, 5987. 173. Hoar, P. E., Kerrick, W. G . L., and Cassidy, P. S. (1979). Science 204, 503. 174. Peterson, J . W., Ill (1982). Biophys. J . 37, 453. 175. Aksoy, M. 0.. Mras, S., Kamm, K. E., and Murphy, R. A. (1983). Am. J . Physiol. 245, C255. 176. Pemrick, S. M. (1980). JBC 255, 8836. 177. Manning, D. R., and Stull, I. T. (1979). BBRC 90, 164. 178. Manning, D. R., and Stull, J. T. (1982). Am. 1. Physiol. 242, C234.

164

J. T. STULL, M. H. NUNNALLY, ANDC. H. MICHNOFF

Klug, G . A,, Botterman, B. R., and Stull, J . T. (1982). JBC 257, 4688. Houston, M. E., Green, H. J., and Stull, J. T. (1985). Pfluegers Arch. 403, 348. Persechini, A., Stull, J. T., and Cooke, R. (1985). JBC 260, 7951. Nairn, A. C., Hemmings, H. C., Jr., andGreengard, P. (1985). Annu. Rev. Biochem. 54,931. Ahmad, Z., DePaoli-Roach, A. A., and Roach, P. J. (1982). JBC 257, 8348. Payne, M. E., Schworer, C. M., and Soderling, T. R. (1983). JBC 258, 2376. Schworer, C. M., Payne, M. E., Williams, A. T., and Soderling, T. R. (1983). ABB 224,77. Woodgett, J. R., Davison, M. T., and Cohen, P. (1983). EJB 136, 481. Kennedy, M. B., McGuinness, T., and Greengard, P. (1983). J . Neurosci. 3, 818. Kennedy, M. B., Bennett, M. K., and Erondu, N. E. (1983). PNAS 80,7357. Bennett, M. K., Erondu. N. E., and Kennedy, M. B. (1983). JBC 258, 12735. Palfrey, H. C., Rothlein, J. E., and Greengard, P. (1983). JBC 258, 9496. Palfrey, H. C., Lai, Y., and Greengard, P. (1984). In “The Red Cell: Sixth Ann Arbor Conference” (G. J. Brewer, ed.), p. 291. A. R. Liss, Inc., New York. 192. Palfrey, H. C. (1984). FP 43, 1466 (abstr.). 193. Gorelick, F. S . , Cohn, J. A., Freedman, S. D., Delahunt, N. G., Gershoni, J. M., and Jamieson, J. D. (1983). J. Cell Biol. 97, 1294. 194. DeRiemer, S. A., Kaczmarek, L. K., Lai, Y., McGuinness, T. L., and Greengard, P. (1984). J . Neurosci. 4, 1618. 195. Fukunaga, K., Yamamoto, H., Matsui, K., Higashi, K.. and Miyamoto, E. (1982). J. Neurochem. 39, 1607. 196. Goldenring, J. R., Gonzalez, B., McGuire, J . S., Jr., and DeLorenzo, R. J . (1983).JBC 258, 12632. 197. Schworer, C. M., McClure, R. W., and Soderling, T. R. (1983). FP 43, 1466 (abstr.). 198. Kuret, J . , and Schulman, H. (1984). Biochemistry 23, 5495. 199. Yamauchi, T., and Fujisawa, H. (1983). EJB 132, 15. 200. Sahyoun, N., LeVine, H., 111, Bronson, D., Siegel-Greenstein, F., and Cuatrecasas, P. (1985). JBC 260, 1230. 201. McGuinness, T. L., Lai, Y., and Greengard, P. (1985). JBC 260, 1696. 202. Goldenring, J. R., Casanova, J. E., and DeLorenzo, R. I. (1984). J . Neurochem. 43, 1669. 203. Kelly, P. T., McGuinness, T. L., and Greengard, P. (1984). PNAS 81, 945. 204. Sahyoun, N., LeVine, H., 111, and Cuatrecasas, P. (1984). PNAS 81, 431 I . 205. McGuinness, T. L., Lai, Y., Greengard, P., Woodgett, J. R., and Cohen, P. (1983). FEES Lett. 163, 329. 206. Woodgett, J. R., Cohen, P., Yamauchi, T., and Fujisawa, H. (1984). FEES Lett. 170, 49. 207. Kloepper, R. F., and Landt, M. (1984). Cell Calcium 5, 351. 208. Laemmli, U. K. (1970). Nature (London) 227, 680. 209. Cohn, J. A., Gorelick, F. S., Delahunt, N. G . , and Jamieson, J . D. (1984). FP 43, 1466 (abstr.). 210. Walaas, S. I . , Nairn, A. C., and Greengard, P. (1983). J. Neurosci. 3, 291. 211. Walaas, S . I., Nairn, A. C., and Greengard, P. (1983). J. Neurosci. 3, 302. 212. Ouimet, C. C., McGuinness, T. L., and Greengard, P. (1984). PNAS 81, 5604. 213. Schworer, C. M., and Soderling, T. R. (1983). BBRC 116, 412. 214. Doskeland, A. P., Schworer, C. M., Doskeland, S. O., Chrisman, T. D., Soderling, T. R., Corbin, J. D., and Flatmark, T. (1984). EJB 145, 31. 215. Wise, B. C., Guidotti, A,, and Costa, E. (1984).Adv. Cyclic Nucleoridefrotein Phosphorylution Res. 17, 5 I I . 216. Schworer, C. M., El-Maghrabi, M. R., Pilkis, S. J., and Soderling, T. M. (1985). FP 44,705 (abstr.). 217. Sahyoun, N., LeVine, H., 111, Bronson, D., and Cuatrecasas, P. (1984). JBC 259, 9341. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191.


165

218. Camici, M., Ahmad, Z., DePaoli-Roach, A. A , , and Roach, P. J. (1984). JBC 259, 2466. 219. Sobue, K., Kanda, K., and Kakiuchi, S. (1982). FEBS Lett. 150, 185. 220. Iwasa, T., Fukunaga, K., Yamamoto, H., Tanaka, E., and Miyamoto, E. (1984). ABB 235, 212. 221. Penn, E. J . , Brocklehurst, K. W., Sopwith, A. M., Hales, C. N., and Hutton, J . C. (1982). FEES Lett. 139, 4. 222. Krebs, E. G . , Love, D.S., Bratvold, 0 . E., Trayser, K. A,, Meyer, W. L., and Fischer, E. H. (1964). Biochemistry 3, 1022. 223. Tabatabai, L. B., and Graves, D. J . (1978). JBC 253, 2196. 224. Yamamura, H., Nishiyama, K., Shimomura, R.. and Nishizuka, Y. (1973). Biochemistry 12, 856. 225. Sugden, P. H., Holladay, L. A., Reimann, E. M., and Corbin, J. D. (1976). BJ 159, 409. 226. Inoue, M., Kishimoto, A., Takai, Y., and Nishizuka, Y. (1976). JBC 251, 4476. 227. Glass, D. B . , McFann, L. J., Miller, M. D., and Zeilig, C. E. (1981). In “Protein Phosphorylation” (0.M. Rosen and E. G. Krebs, eds.), Vol. 8, p. 267. Cold Spring Harbor Press, Cold Spring Harbor, New York. 228. Dahmus, M. E. (1976). Biochemistry 15, 1821. 229. Dahmus, M. E. (1981). JBC 256, 3319. 230. Dahmus, M. E., and Natzle, J . (1977). Biochernisrry 16, 1901. 231. Hathaway, G. M., and Traugh, J. A. (1979). JBC 254, 762. 232. Yamamoto, H . , Fukunaga, K., Goto, S., Tanaka, E., and Miyamoto, E. (1985). J . Neurochem. 44, 759. 233. Schulman, H. (1984). J. Cell B i d . 99, I I . 234. Fujisawa, H., Yamauchi, T., Nakata, H., and Okuno, S. (1984). FP 43, 3011. 235. Fukunaga, K., Yamamoto, H., Tanaka, E., Iwasa, T., and Miyamoto, E. (1984).Life Sci. 35, 493. 236. Vulliet, P. R., Woodgett, J . R., and Cohen, P. (1984). JBC 259, 13680. 237. Ueda, T., Maeno, H., and Greengard, P. (1973). JBC 248, 8295. 238. Huttner, W. B., and Greengard, P. (1979). PNAS 76, 5402. 239. Huttner, W. B., DeGennaro, L. J . , and Greengard, P. (1981). JBC 256, 1482. 240. Embi, N., Parker, P. J., and Cohen, P. (1981). EJB 115, 405. 241. Kemp, B . E., Pearson, R. B., Woodgett, J. R., and Cohen, P. (1985). FP 44, 705 (abstr.). 242. Hofmann, F., Beavo, J. A., Bechtel, P. J . , and Krebs, E. G. (1975). JBC 250, 7795. 243. Rangel-Aldao, R . , Kupiec, J. W., and Rosen, 0. M. (1979). JBC 254, 2499. 244. Scott, C. W., and Mumby, M. C. (1985). JBC 260, 2274. 245. Hemmings, B . A., Yellowlees, D., Kemohan, J . C., and Cohen, P. C. (1981). EJB 119,443. 246. Shields, S. M., Vernon, P. J., and Kelly, P. T. (1984). J . Neurochem. 43, 1599. 247. Zoller, M. J., and Taylor, S. S. (1979). JBC 254, 8363. 248. LeVine, H., 111, Sahyoun, N. E., and Cuatrecasas, P. (1985). PNAS 82, 287. 249. Kennedy, M. B . (1983). Annu. Rev. Neurosci. 6, 493. 250. Schulman, H. (1984). Trends Pharmacol. Sci. 5, 188. 251. Nestler, E. J . , Walaas, S. I., and Greengard, P. (1984). Science 225, 1357. 252. Pearson, R. B . , Jakes, R., John, M., Kendrick-Jones, J., and Kemp, B. E. (1984).FEBSLett. 168, 108. 253. Suzuyama, Y . , Umegane, T., Maita, T., and Matsuda, G . (1980). Hoppe-Seyler’sZ. Physiol. Chem. 361, 119. 254. Collins, J . H. (1976). Narure (London) 259, 699. 255. Matsuda, G., Maita, T., Suzuyama, Y., Setoguchi, M., and Umegane, T. (1977). J . Biochem. (Toyko) 81, 809. 256. Matsuda, G., Maita, T., Kato, Y., Chen, J.-I., and Umegane, T. (1981). FEBSLett. 135,232.

166


257. 258. 259. 260. 261. 262. 263.

Burke, 9. E., and DeLorenzo, R. J. (1981). PNAS 78, 991. Huang, T. S., and Krebs, E. G. (1979). FEES Lerr. 98, 66. Rylatt, D. B., and Cohen, P. (1979). FEES Len. 98, 71. Huang, T. S . , and Krebs, E. G. (1977). BBRC 75, 643. Parker, P. J., Aitken, A., Bilham, T., Embi, N.. and Cohen, P. (1981). FEESLerr. 123,332. Wretbom, M., Humble, E., Ragnarsson, U., and Engstrom, L. (1980). BBRC 93, 403. Hjelmquist, G., Anderson, J., Edlund, B., and Engstriim, L. (1974). EBRC 61, 509.

Protein Kinase C USHIO KIKKAWA

.

YASUTOMI NISHIZUKA

Department of Biochemistry Kobe University School of Medicine Kobe 650, Japan

I. Introduction

.............................................

11. Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Biochemical Activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Physiological Activation . ..............

Action of Tumor Promoters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inhibitors . . . . . . . . . . . . . . . . . . ............ Synergistic Roles with Calcium ............ Growth Response and Down Reg ............................. Target Proteins and Catalytic Specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relation to Other Receptors ........................ XI. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . ..........

V. VI. VII. VIII. IX. X.

1.

167 168 169 171

173 174 175 177 179 181 183 183

Introduction

The biochemical basis of signal transduction across the cell membrane has long been a subject of interest, and protein kinase C has attracted great attention in the studies on the control of cellular functions and proliferation. Under physiological conditions the enzyme is activated by diacylglycerol in the presence of Ca2+ and membrane phospholipid. The diacylglycerol active in this role may arise in the plasma membrane only transiently from the receptor-mediated hydrolysis of inositol phospholipids, and this protein kinase appears to be indispensable for transmitting various extracellular informational signals from the cell surface into the cell interior. The signals relating to protein kinase C normally I67 THE ENZYMES. Vol. XVlI Copyrighl 0 1986 by Academic Press. Inc. All rights of reproduction in any Form reserved.

168

USHlO KIKKAWA AND YASUTOMI NISHIZUKA

differ from those from a group of hormones and neurotransmitters that produce cyclic AMP as an intracellular messenger. Thus, protein kinase C and cyclic AMP-dependent protein kinase (protein kinase A) transduce distinctly different pieces of information into the cell through their own specific protein phosphorylation. This article describes some properties, mechanism of activation, and possible roles of protein kinase C in cell-surface signal transduction. Several aspects of this protein kinase system have been reviewed ( 1-6).

II. Properties Protein kinase C was first found in 1977 as a proteolytically activated protein kinase that was capable of phosphorylating histone (7), and later found to be activated reversibly by association of membrane phospholipid in the presence of diacylglycerol and physiological concentration of Ca2 (8, 9). This enzyme is ubiquitously distributed in tissues and organs, with the brain having highest activity (10, 11). In this tissue a large quantity of the enzyme is associated with synaptic membranes, whereas in most other tissues the enzyme is present mainly in the soluble fraction as an inactive form and recovered upon biochemical fractionation of subcellular components (12). The enzyme has been purified to near homogeneity from the soluble fraction of rat brain (12), bovine heart (13), pig spleen ( 1 4 , bovine brain (15), rabbit renal cortex (16),and rabbit brain (17). The original method for purification from rat brain was later improved (18). The brain enzyme shows a single band upon sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis with a molecular weight of about 82,000. The Stokes radius is 42& and the molecular weight is 87,000 as estimated by gel filtration analysis. The sedimentation coefficient is 5.1 S which corresponds to a molecular weight of 77,000. The frictional ratio of the enzyme is calculated to be 1.6, indicating an asymmetric nature of the molecule. Although the molecular weight mentioned above slightly varies with the methods employed for estimation, the enzyme is composed of a single polypeptide chain with no subunit structure. Neither calmodulin nor an antibody against calmodulin affects the enzymic activity. The isoelectric point of the enzyme is pH 5.6. The optimum pH range for activity is 7.5-8.0 with Tris acetate as a test buffer. Mg2+ is essential for the catalytic activity with the optimum range having about 5- 10 mM. To some extent Mg2+ can be replaced by Mn2 or Co2 . The optimum concentrations for Mn2 and Co2 are 0.5- 1 mM, with the maximum enzymic activity being approximately 50% of that with Mg2 . The K,,, value for ATP is about 6 X 10W6M. Guanosine triphosphate does not serve as a phosphate donor. Protein kinase C utilizes ATP-y-S as a phosphoryl donor, but the reaction velocity is very slow (19). The enzymes obtained from various tissues appear to be similar and practically +

+

+

+

+

+

5.

169

PROTEIN KINASE C

indistinguishable from one another at least in their kinetic and catalytic properties. However, there are minor differences; for example, the molecular weight of the heart enzyme is about 83,500 on SDS-polyacrylamide gel electrophoresis, 99,500 on sedimentation coefficient (5.6 S) and Stokes radius (42.9 A), and 113,000 on gel filtration analysis (13).The optimum pH of heart enzyme is 6-7 with Pipes as a test buffer. It is not known at present whether these differences are due to intrinsic properties of the enzymes from different species and different tissues. Some heterogeneity of protein kinase C in brain and other tissues has been described, and the enzyme is very susceptible to proteolysis (7, 12).

111.

Biochemical Activation

Protein kinase C per se is normally inactive. When assayed in a cell-free enzymic reaction, the protein kinase depends on Ca2+ as well as phospholipid for its activation. However, diacylglycerol, which is produced in membranes from inositol phospholipids in a signal-dependent fashion, dramatically increases the affinity of this enzyme to Ca2+, and thereby renders it fully active without a net increase in the Ca2+ concentration (9, 2 0 ) . Thus, the activation of this unique protein kinase is biochemically dependent on, but physiologically independent of, Ca2+. Figure 1 shows some kinetic analyses indicating that the sensitivity of protein kinase C to this divalent cation is greatly increased by the addition of diacylglycerol. Among various phospholipids tested, phosphatidylserine is absolutely required for the enzyme activation. At lower concentrations of Ca2 other phospholipids such as phosphatidylethanolamine,phosphatidylinositol, phosphatidylcholine, and sphingomyelin are all inert. However, several of these phospholipids show positive or negative cooperativity for the activation of protein kinase C when supplemented to phosphatidylserine. For instance, phosphatidylethanolamine increases further the affinity of enzyme for Ca2 , and makes it fully active at the range of lo-’ M of this cation. whereas both phosphatidylcholine and sphingomyelin show opposing effects (20). Thus, the asymmetric distribution of various phospholipids in the membrane phospholipid bilayer may take a part in the activation of enzyme, and the K , value for Ca2+ depends on the phospholipid composition as well as on the presence of diacylglycerol. The enzyme activation is specific to Ca2 , and none of other divalent cations tested is able to substitute for Ca2 , except Sr2 which is about 10% as active as Ca2+ at comparable concentrations. In the enzymic reaction various diacylglycerols are capable of activating the enzyme. In physiological processes, 1-stearoyl-2-arachidonyl glycerol is likely the activator of this enzyme, since most of inositol phospholipids contain this diacylglycerol backbone (21). It was initially found that diacylglycerol active in +

+

+

+

+

170

USHIO KIKKAWA AND YASUTOMI NISHIZUKA

//

lo-’

Phospholipid alone

lo-’

CaC1,

(MI

and phospholipid. Homogeneous protein kinase C obtained from rat brain was assayed in the presence of various concentrations of Ca2 by measuring the incorporation of the radioactive phosphate of (y-32PIATPinto calf thymus HI histone as a model phosphate acceptor. Where indicated, diolein and a mixture of phospholipids from human erythrocytes were added. Other detailed conditions are described elsewhere (9) [taken from Ref. ( 5 ) ] . FIG.

1. Activation c. >rotein kinase C by diacylglycerol in the presence of Ct +

this role contains at least one unsaturated fatty acyl moiety at either 1 or 2 position (9). However, it became evident that when one fatty acyl moiety is replaced by a short chain the resulting diacylglycerols such as 1 -palmitoyl-2acetyl glycerol (22), dioctanoyl glycerol and dihexanoyl glycerol (23, 24) are also active to support the enzyme activation. More recently, the active diacylglycerol is found to be specific to the 1,2-sn-configuration, and other stereoisomers are totally inactive, suggesting that a highly specific lipid-protein interaction is needed for this enzyme activation (25). However, at higher concentrations of Ca2+, the enzyme exhibits catalytic activity in the presence of phospholipid but without diacylglycerol as noted above. The detailed biochemical mechanism of this enzyme activation remains largely unknown. Protein kinase C is alternatively activated by proteolysis with Ca2 -dependent protease or trypsin (7, 26). When Ca2+-dependent protease is employed, a limited proteolysis takes place, and a smaller component carrying enzymic activity is produced. The molecular weight is estimated to be about 5 1,000. The catalytically active component thus produced is totally independent of Ca2+ , +

171

5. PROTEIN KINASE C

phospholipid, and diacylglycerol. Protein kinase C, which is attached to the membrane, is more susceptible to limited proteolysis by Ca2+-dependent protease (26). The protein kinases fully activated either proteolytically or nonproteolytically show similar kinetic and catalytic properties, and exhibit the same levels of enzymic activity. Although it has been proposed that such proteolytic activation of protein kinase C may occur in intact cell systems ( 2 3 , the physiological significance of this proteolysis has not been defined.

IV. Physiological Activation A wide variety of hormones, neurotransmitters, and many other biologically active substances activate cellular functions and proliferation through interaction with their specific cell surface receptors. It has been repeatedly shown that many of these signals provoke the breakdown of inositol phospholipids in the plasma membrane (28-31). Initially, phosphatidylinositol (PI) was regarded as a prime target (32), but further evidence seems to suggest that, after stimulation of the receptor, phosphatidylinositol-4,5-bisphosphate(PIP,) rather than PI and phosphatidylinositol-4-phosphate(PIP) is degraded immediately to produce 1,2-diacylglycerol and inositol 1,4,5-trisphosphate (IP,) (33-37). In general, this turnover of membrane phospholipids is associated with an increase in intracellular concentration of Ca2+, which appears to mediate many of the subsequent physiological responses. Studies in this laboratory have provided evidence that the diacylglycerol produced in this way initiates the activation of protein kinase C, and thereby the information of extracellular signals is translated to protein phosphorylation as shown in Fig. 2. It has been proposed that IP,, the other product of PIP, breakdown, could serve as a mediator of Ca2 release from intracellular stores (38).Evidence for this proposal has been obtained from the studies on the effect of IF, on various permeabilized cells, and this Ca2 -releasing activity of IP, was first demonstrated in a preparation of rat pancreatic acinar cells (39). This possibility has been subsequently supported using several permeabilization and Ca2 -measuring procedures in many other cell types and also using micro-injection of IP, into photoreceptors (38). If this proposal is correct, then the signal-dependent breakdown of a single molecule, PIP,, may generate two intracellular mediators, diacylglycerol that induces protein kinase C activation and IP, that mobilizes Ca2+. However, it is still not absolutely clear that only PIP, is hydrolyzed in response to external stimuli. It is possible that three inositol phospholipids are broken down by phospholipases C at different times at different rates or that these phospholipids are hydrolyzed at different sites within the cell (6). The phospholipases C in mammalian tissues, except for the enzyme of lysosomal origin, normally require high concentrations of Ca2 when assayed in cell-free +

+

+

+

172


Extracellular Signal

PI

PIP

PI P2

Receptor

PhYslo 1og 1 ca 1 Responses

FIG. 2. Inositol phospholipid turnover and cell-surface signal transduction. PI, phosphatidylinositol; PIP, phosphatidylinositol-4-phosphate; PIP2, phosphatidylinositol-4,5-bisphosphate; CDP.DG, CDP-diacylglycerol; PA, phosphatidic acid; DG, 1,2-diacylglycerol; IP3, inositol-l,4,5-triphosphate;I, inositol moiety; and P, phosphoryl group [adapted from Ref. (31.

systems. Although the regulatory mechanism of this enzyme is not clear, various results suggest that the breakdown of inositol phospholipids is Ca2 -dependent but may not be regulated by Ca2+ (37). Evidence suggests that GTP may be involved in this signal-induced hydrolysis of inositol phospholipids (40-43). Although it is generally accepted that the inositol phospholipid turnover is firmly linked to the activation of protein kinase C, the evidence for this signal transduction has primarily come from experiments with platelets as a model system, where many agonists and antagonists for the aggregation and release reaction are known (44,45).When stimulated by thrombin, collagen, or plateletactivating factor (PAF), two endogenous platelet proteins with approximate molecular weights of 47,000 (47K protein) and 20,000 (20K protein) are heavily phosphorylated, and this phosphorylation reaction is normally associated with the release of their constituents such as serotonin (46, 47). The 20K protein is myosin light chain, and a calmodulin-dependent protein kinase is responsible for this phosphorylation reaction (48).Although the function of 47K protein remains unknown, the phosphorylation of this protein may be used as a marker of the activation of protein kinase C (45). When human platelets are stimulated, diacylglycerol containing arachidonate is rapidly produced with the concomitant phosphorylation of 47K protein (44, 45, 49, 50). +

5.

173

PROTEIN KINASE C

Under appropriate conditions synthetic diacylglycerol such as 1-0leoy1-2-acetyl glycerol is permeable and directly activates protein kinase C without interaction with cell surface receptors (51, 5 2 ) . This exogenously added diacylglycerol does not produce endogenous diacylglycerol nor does it induce inositol phospholipid breakdown. There is no indication of arachidonate release or damage of cell membranes. Instead, the exogenous diacylglycerol is rapidly converted in situ to the corresponding phosphatidate, that is, 1-oleoyl-2-acetyl-3-phosphoryl glycerol, probably through the action of diacylglycerol kinase. Furthermore, it has been shown that dioctanoyl glycerol and dihexanoyl glycerol are also permeable to cell membranes and activate protein kinase C directly ( 2 3 , 2 4 ) .Several lines of experimental evidence seem to indicate that protein kinase C is activated upon stimulation by various extracellular signals, and that the receptor-mediated hydrolysis of inositol phospholipids is a sign for the transmembrane control of protein phosphorylation. In this process protein kinase C may be reversibly attached to membranes, and presumably a quaternary complex consisting of the enzyme, phospholipid, diacylglycerol, and Ca2 is produced. Diacylglycerol and Ca2+ may be synergistically effective to produce such a catalytically active complex, but the precise physiological picture of this unique lipid-protein interaction remains to be clarified. +

V. Action of Tumor Promoters Tumor-promoting phorbol esters, such as 12-0-tetradecanoyl-phorbol-13acetate (TPA) first isolated from croton oil, elicit a variety of biological and biochemical actions in a manner very similar to hormones. In most cases Ca2 is indispensable for causing such cellular responses, and a number of kinetic studies with various cell types suggest that their primary site of action is the cell surface membrane (53-57). Evidence is available that protein kinase C is a target for phorbol esters, since the tumor promoters directly activate this enzyme both in vitro and in vivo, and there is an approximate correlation between the ability of individual phorbol esters to promote tumor development and to activate the protein kinase (58, 59). Kinetic analysis indicates that TPA, which has a diacylglycerol-like structure, is able to substitute for diacylglycerol at extremely low concentrations. Like diacylglycerol, TPA dramatically increases the affinity of the enzyme for Ca2+ to the l o p 7 M range, resulting in its full activation without detectable mobilization of Ca2 when measured by an intracellular Ca2 -indicator, quin 2. Studies with a homogeneous preparation of protein kinase C indicate that [3H]phorbol-12,13-dibutyrate (PDBu), which is another potent tumor-promoting phorbol ester, may bind to the enzyme only when Ca2+ and phospholipid are present (60). This radioactive phorbol ester binds neither to protein kinase C nor to phospholipid per se irrespective of the presence or absence of Ca2+, and all +

+

+

174


four components previously mentioned are needed simultaneously for the binding as well as for the enzyme activation. The apparent dissociation binding constant (K,) of the tumor promoter is exactly identical with the activation constant (KO)for the enzyme, and this value varies with the composition of phospholipids added to the incubation mixture. Again, phosphatidylserine is essential, and other phospholipids show positive or negative cooperativity for the binding. This may help explain the reported existence of apparent multiple binding sites of tumor promoter in broken cell preparations (61). In any case, the Kd values obtained with purified protein kinase C are remarkably similar to those previously described for the specific tumor-promoter-binding site on intact cell membranes (62-65). Diacylglycerols such as diolein compete with the radioactive phorbol ester for the binding, whereas neither monoolein, triolein, nor free oleic acid is active in this capacity under similar conditions. Scatchard analysis indicates that roughly one molecule of PDBu binds to one molecule of protein kinase C in the presence of a physiological concentration of Ca2+ and an apparent excess of phospholipid (60). Presumably in intact cells, where phospholipid and Ca2+ are not limited, for each molecule of tumorpromoting phorbol ester intercalated into the membrane phospholipid bilayer, one molecule of protein kinase C moves to it and produces the quaternary complex by which the enzyme is activated. Studies in this and other laboratories (60-69) strongly suggest that protein kinase C is a receptor protein of tumorpromoting phorbol esters, and that many of the pleiotropic actions, if not all, of the tumor promoters may be mediated through the action of protein kinase C. It may be noted that mezerein (70), teleocidin, and debromoaplysiatoxin (71), which have no diacylglycerol-like structure but exhibit tumor-promoting activity, are all capable of activating protein kinase C presumably by causing membrane perturbation analogous to that phorbol esters do. However, it is still possible that these tumor promoters have additional actions on the membrane, particularly at higher concentrations. For instance, TPA may act as a weak Ca2 -ionophore under certain conditions. +

VI. Inhibitors No inhibitor has been found that is specific for protein kinase C. Theoretically, at least three entities of inhibitors may exist: first, compounds that antagonize the action of diacylglycerol; second, phospholipid-interacting compounds that prevent the activation of the enzyme; and third, compounds that inhibit the catalytically active center of the enzyme. It is obviously important to develop specific inhibitors of this enzyme for several practical reasons. However, no inhibitor has been found that belongs to the first entity. Instead, R 59 022 (6-[2-[4-[(4-fluorophenyl)phenyImethylene] - 1- piperidinyllethyl] -7- methyl-5H- thiazolo[ 3,2-a]

5.

175

PROTEIN KINASE C

pyrimidin-5-one) is shown to act as a selective inhibitor of diacylglycerol kinase, and thereby enhances the accumulation of diacylglycerol in membranes (72). Most of the inhibitory compounds described appear to be in the second entity. For instance, protein kinase C is inhibited to various extents by psychotic drugs (trifluoperazine, chlorpromazine, fluphenazine, imipramine, etc.), local anesthetics (dibucaine, tetracaine, etc.), W-7 [N(6-aminohexyl)-5-chloro-1naphthalenesulfonamide], verapamil, phentolamine, adriamycin, polyamines (spermine, spermidine, and putrescine), palmitoylcarnitine, melittin, heparin, polymixin B, and vitamin E (10, 73-77). The inhibition of protein kinase C by these drugs is not due to their interaction with the active center of the enzyme, since the catalytically active enzyme fragment, which is obtained by limited proteolysis (26),is not susceptible to any of these drugs. Kinetically, most of the drugs listed above interact with phospholipid, and inhibit the activation of the enzyme in a competitive manner. These phospholipid-interacting drugs usually also inhibit calmodulin-dependent protein kinases such as myosin light chain kinase by competing with calmodulin. In an experiment using intact platelets it was shown that some of these drugs such as chlorpromazine, dibucaine, and tetracaine do not inhibit thrombin-induced diacylglycerol formation, but in fact profoundly inhibit the activation of protein kinase C in a dose-dependent manner (44, 45). Cyclic nucleotide-dependent protein kinases are not susceptible to these drugs. Some inhibitors that interact with the catalytically active center of protein kinases have been described (78). For instance, H-7 [ 1-(5-isoquinolinesulfonyl)-2-methylpiperazine] profoundly inhibits protein kinase C. However, protein kinase A is inhibited by this compound as well, but myosin light chain kinase is far less susceptible. It has been reported that some polypeptide cytotoxins effectively and specifically inhibit protein kinase C relative to protein kinase A and myosin light chain kinase, but the mode of their inhibitor actions is not known (79).

VII. Synergistic Roles with Calcium Although bovine adrenal medullary cells (80) and some presynaptic muscarinic receptors (81) may be exceptions, it is generally the case that when stimulation of receptors leads to inositol phospholipid breakdown it simultaneously mobilizes Ca2+. A series of studies indicate that the activation of protein kinase C appears to be a prerequisite but not a sufficient requirement for physiological responses of target cells, because the cellular responses to synthetic diacylglycerol or to tumor promoter per se are always incomplete. Under appropriate conditions it is possible to induce protein kinase C activation and Ca2 mobilization selectively by the addition of permeable diacylglycerol or tumor promoter for the former and a Ca2 -ionophore, such as A23 187, for the +

+

176


latter. By using this procedure it is possible to demonstrate that protein kinase C activation and Ca2 mobilization are both essential and act synergistically to elicit full physiological responses such as release reaction of serotonin (51, 52). In this experiment the concentration of A23 187 (0.2-0.4 pM) is critical, because at higher concentrations of more than 0.5 pll this Ca2+-ionophore itself will cause the phosphorylation of 47K protein as well as 20K protein, probably due to the nonspecific activation of phospholipase C that is accompanied by a large increase in Ca2 concentration. Likewise, the synthetic diacylglycerol or tumor promoter alone at higher concentrations (more than 50 pg/ml or 50 ng/ml, respectively) causes a significant release of platelet constituents. The precise reason for this enhanced release is unclear, but it is possible that these compounds can induce the release reaction by acting as membrane fusigens or weak Ca2 -ionophores. The involvement of the two pathways, protein kinase C activation and Ca2+ mobilization, in the signal transduction may explain, at least in part, the agonistselectivity that is often observed in release reactions. For instance, again in platelets, serotonin and adenine nucleotides are released from dense bodies in response to a variety of signals such as thrombin, collagen, ADP, epinephrine, and PAF, while lysosomal enzymes are released only at higher concentrations of thrombin and collagen. By using permeabilized platelets it has been shown that such agonist-selectivity of release reactions is not related to Ca2 concentrations, because there is no difference in their sensitivity to Ca2+ (82).Theoretically, it is possible that the two pathways mentioned may exert differential control over release reactions from different granules within a single activated platelet. In neuronal tissues, a single nerve ending frequently contains both peptides and classical transmitters presumably present in different stores (83). It is possible to imagine that the two synergistic routes may also be responsible for the frequencyselectivity of release reactions that is often observed during electrical stimulation. It is well known that depolarization of membranes by electrical stimulation or depolarizing agents induces inositol phospholipid turnover (84-87). A potential role of protein kinase C for the activation of cellular functions has been suggested in many other systems. For instance, the synergistic roles of this enzyme with Ca2 are proposed for the receptor-mediated release reactions and exocytosis of several endocrine as well as exocrine tissues as exemplified in Table I (88-109). It seems important to note that in both peripheral and central nervous tissues the two pathways appear to be essential for the neurotransmitter release from nerve endings. In fact, the combination of TPA and A23 187 has been shown to induce full activation of acetylcholine release from the cholinergic nerve endings of guinea pig ileum (101).The physiological responses through these two pathways may include not only release reactions of various cell types but also many other cellular processes such as adrenal steroidogenesis (102), neutrophil superoxide generation (103-105), smooth muscle contraction (106, 1 0 3 , and hepatic glycogenolysis (108, 109). +

+

+

+

+

5.

177

PROTEIN KINASE C TABLE I CELLULAR RESP~NSES PROBABLY ELICITEDB Y SYNERGISTIC ACTIONSOF PROTEINKINASEC AND C A ~ + Tissues

Responses

References

1. Release Reactions and Exocytosis

Platelets

Serotonin Lysosomal enzyme Mast cells Histamine Lysosomal enzymes Neutrophils Adrenal medulla Catecholamine Adrenal cortex Aldosterone Pancreatic islets Insulin Pancreatic acini Amylase Pituitary cells Gonadotropin Th yrotropin Ileal nerve endings Acetylcholine 2. Metabolic Processes-and Others Adrenal cortex Steroidogenesis Neutrophils Superoxide generation Smooth muscle Contraction Hepatocytes GIycogenol ysis

(51,52,88) (88,89) (90) (89,91,92) (93)

(102) (103-105) (106.107) (108.109)

Alternatively, it has been proposed that the activation of protein kinase Cper se is sufficient to induce some cellular responses such as serotonin release from platelets (110) and superoxide generation and exocytosis of neutrophils (ZZZ), and these reactions are shown to proceed without any detectable increase in Ca2+ concentrations only when protein kinase C is activated. The precise relationship between protein kinase C and Ca2+ actions during the activation of various cellular functions is a subject of great interest.

VIII. Growth Response and Down Regulation The role of two pathways, protein kinase C activation and Ca2 mobilization, is not only confined to the short-term responses described earlier in this chapter but is also extended to the long-term responses such as cell proliferation. It is possible to show with macrophage-depleted human peripheral lymphocytes that the two pathways are both essential and synergistically act for promoting DNA synthesis (112, 113). However, for long-term responses, a low concentration of some growth factor such as phytohemagglutinin is still needed, implying that another hitherto unknown signal pathway is involved in eliciting full activation of cell proliferation (112). In a similar set of experiments with Swiss 3T3 cell line, insulin is necessary for the growth response in addition to synthetic diacylglycerol or TPA +

178


(Growth factors) (Tyrosine kinases 3 ) (Internalization ? j FIG.3. Pathways of signal transduction for short-term and long-term cellular responses. DG, 1,2diacylglycerol; and IP3, inositol-l,4,5-trisphosphate.

(114). It has been known for some time that tumor promoter and growth factor such as epidermal growth factor (EGF) generally act in concert for cell proliferation (115). Platelet-derived growth factor (PDGF) or TPA is shown to induce messenger RNAs of some protooncogenes such as c-fos within 5-10 min (116), although the concentration of TPA employed for this experiment was extremely high. Figure 3 illustrates the hypothetical pathways of signal transduction for short-term and long-term cellular responses. Some of the receptors for growth factors are associated with a tyrosine-specificprotein kinase activity. The role of protein kinase C and tyrosine-specific protein kinases in cell proliferation is another subject of current interest. Very frequently protein kinase C appears to be related to “down regulation” or “negative feedback control.” For instance, protein kinase C is shown to phosphorylate the receptor protein of some growth factors such as EGF with the concomitant decrease in both its tyrosine-specific protein kinase and growth factor-binding activities (24, 117-122). In EGF receptor, threonine-654 is shown to be phosphorylated by protein kinase C (123), which is located at the upstream close to its tyrosine-specific protein kinase domain (124). Although TPA markedly reduces EGF-binding to many mitogenically responsive cell types (125-128), the biological significance of this phosphorylation reaction remains to be explored. In an analogous fashion, it has been suggested that some receptors in human leukemic cell lines, such as the receptors of insulin (129), somatomedin C (1291, transferrin (130, 131), and interleukin 2 (132), and also a,-adrenergic receptor in rat hepatocytes (133),are phosphorylated by protein kinase C. However, a logical consequence of the phosphorylation reactions of these receptors has not been well substantiated. The dual functions of protein kinase C described in the preceding section, apparently positive forward action and negative feedback action, are observed also in some other cellular processes. Myosin light chain is shown to be phosphorylated

5.

179

PROTEIN KINASE C

by protein kinase C, and the site of phosphorylation differs from that phosphorylated by myosin light chain kinase (134). TPA and Ca2+-ionophore are synergistically involved in smooth muscle contraction (106, 107), but the phosphorylation of myosin light chain by protein kinase C appears to suppress the actin-activated ATPase activity of myosin, resulting in the relaxation (135). It is possible that this reaction catalyzed by protein kinase C may constitute a feedback mechanism to prevent overresponse, since the reaction proceeds slowly compared with the contraction, which is completed within seconds. A similar negative feedback control by protein kinase C may be possible for the signal-induced increase of Ca2+ (110, 136-138). Such a mechanism to decrease the concentration of Ca2 has been described for Ca2 -transport ATPase, which is activated by this divalent cation through a calmodulin-dependent mechanism (139, 140). Analogously, in cardiac systems, protein kinase C phosphorylates sarcoplasmic reticulum proteins, resulting in the enhanced Ca2 -transport ATPase and, thus, leading to the decrease in the cytoplasmic Ca2+ concentration (141-144). The significance of such feedback control of the intracellular Ca2+ concentration is also not known. In physiological processes, both diacylglycerol and Ca2 appear only transiently, and the informational signals pass through the membrane very quickly and elicit subsequent cellular responses. Perhaps, in biological systems, a positive signal may be immediately followed by a negative feedback mechanism. In some tissues with phorbol ester, it is possible that such a negative feedback phase may predominate under certain experimental conditions, where the tumor promoter stays in membranes and keeps protein kinase C active for an unusually long period of time. Some other feedback mechanisms, such as the inhibition of protein kinase C by a calmodulin-dependent system (145) and the inhibition of inositol phospholipid breakdown by protein kinase C (146) have been suggested. +

+

+

+

IX. Target Proteins and Catalytic Specificity Although evidence is accumulating that protein kinase C is at times related to apparently negative feedback control as previously discussed, the enzyme plays positive roles in overall processes to activate many cellular functions and proliferation. In most tissues, however, crucial information for such physiological target proteins is unavailable. The enzyme has a broad substrate specificity in vitru, and phosphorylates seryl and threonyl residues but not tyrosyl residue of endogenous proteins. Protein kinase C is phosphorylated by itself in the simultaneous presence of Ca2 ,phospholipid, and diacylglycerol(12). Approximately 2 mol of phosphate are incorporated into each mole of the enzyme, and both seryl and threonyl residues are phosphorylated. Table 11 (147-1 67) lists some proteins that may serve as phosphate acceptors in v i m . Although this list is expanding very +

180

USHlO KIKKAWA AND YASUTOMI NISHIZUKA TABLE I1 POSSIBLEPHOSPHATEACCEFTOR PROTEINSOF PROTEIN KINASEC Phosphate acceptor proteins Receptor proteins Epidermal growth factor receptor Insulin receptor Somatomedin C receptor Transferrin receptor Interleukin 2 receptor Contractile proteins and cytoskeletous proteins Myosin light chain Troponin T and I Filamin Vinculin Microtubule-associated protein Gap junction proteins Membrane and nuclear proteins Histones and protamine High mobility group proteins Middle T antigen CaZ+ -ATPase and phospholamban Synaptic B-50 (Fl) protein Na+-H + exchange protein Enzymes and other proteins Glycogen phosphorylase kinase Glycogen synthetase Guanylate cyclase Initiation factor 2 @-subunit) Fibrinogen Myelin basic protein Retinoid-binding protein Ribosomal S6 protein

References

( I 17-123) (129) (129) (130,131) (132)

(134,135,147) (148) (149) (149,150) (151) (152-154) (7-9) (155)

(156) (141-144) ( 157.158) (137,159)

rapidly, most of these phosphorylation reactions remain to be explored for the physiological significance. For instance, the 47K protein in platelets described earlier in this chapter appears to be involved in the release reaction, but its definitive role is yet to be clarifed. In neuronal tissues, synapsin I, which is located specifically on the cytoplasmic surface of synaptic fesicles, appears to serve as a substrate for protein kinase A as well as for calmodulin-dependentprotein kinase (168). Apparently, this protein is also phosphorylated by protein kinase C, suggesting that it is one of the possible targets of this proteic kinase and plays some role in release reactions (4).B-50 protein (Fl protein), another brain protein which

5.

PROTEIN KINASE C

181

is associated specifically with presynaptic membranes (169-1 72), serves as a preferable substrate specific to protein kinase C (157, 158). This protein phosphorylation selectively increases after long-term potentiation in the hippocampal neural activity, and has been proposed to be related to the expression of synaptic plasticity (173, 174). It is noted that B-50 protein kinase previously found in brain tissues (175) has been later identified as protein kinase C (176-178). The role of protein kinase C may also be extended to the modulation of membrane conductance, channels and active transport, axoplasmic flow, neurotransmitter biosynthesis, and many other neuronal functions by phosphorylating the proteins involved (4).It has been shown that the activation of endogenous protein kinase C by TPA or the microinjection of this enzyme itself enhances the voltage-sensitive calcium current in bag cell neurons (179). Although protein kinases C and A transduce distinctly different pieces of information into the cell as discussed earlier in this chapter, these two protein kinase pathways sometimes cause apparently similar cellular responses and often potentiate each other. Analysis indicates that protein kinases C and A often share the same phosphate acceptor proteins, even the same seryl and threonyl residues in a single protein molecule for phosphorylation. Extensive work by Krebs and his colleagues and others [reviewed in Ref. (ISO)] have shown that the primary structure of the vicinity of the aminoacyl residue to be phosphorylated is one of the determinant factors for the substrate recognition, and that protein kinase A reacts with the seryl and threonyl residues that are usually located at the downstream close to lysyl or arginyl residue. With some model substrate proteins such as myelin basic protein it is shown that, contrary to protein kinase A, protein kinase C appears to favor the hydroxylamino acids that are located at the upstream close to the basic aminoacyl residue. All seryl and threonyl residues that are phosphorylated commonly by these two enzymes have basic aminoacyl residues at both downstream and upstream locations (181).Although it is not known whether this principle can be extended to many other protein substrates, this approach appears to provide a clue to understand the reason why these protein kinases may show different but sometimes similar functions depending on the structure of substrate protein molecule.

X.

Relation to Other Receptors

In receptor functions there may be dramatic heterogeneity and variations from tissue to tissue, but most tissues possess at least two major classes of receptors in transducing information across the membrane. One class is related to cyclic AMP formation, and the other to the inositol phospholipid turnover. In addition, the stimulation of the latter class of receptors normally releases arachidonic acid, and often increases cyclic GMP. Thus, the succeeding events of protein kinase C

182


activation, Ca2 mobilization, arachidonic acid release, and cyclic GMP formation appear to be integrated in a single receptor cascade. The mode of physiological responses may be roughly divided into two groups. In bidirectional control systems in many tissues such as platelets, neutrophils, lymphocytes, and some neuronal cells, the signals that induce the inositol phospholipid turnover promote the activation of cellular functions, whereas the signals that produce cyclic AMP usually antagonize such activation. For instance, in platelets the signal-induced inositol phospholipid breakdown, diacylglycerol formation, 47K protein phosphorylation, and serotonin release are all blocked concurrently by prostaglandins that increase cyclic AMP as observed by dibutyryl cyclic AMP (2, 44, 45). This inhibitory action of cyclic AMP extends to the mobilization of Ca2+, presumably through the decreased formation of IP, as previously mentioned and through the activation of protein kinase A. Protein kinase A has a potential to decrease cytosolic Ca2+ concentration by phosphorylating the regulatory components of Ca2 -activated ATPase, thereby enhancing its catalytic activity (182).However, the molecular basis of the counteraction of inositol phospholipid breakdown by protein kinase A remains to be explored. It has been sometimes proposed that protein kinase C may inhibit agonist-stimulated adenylate cyclase presumably at the point of regulation by guanine nucleotides (183). However, such an interaction of the two receptor functions has not been unequivocally established. In contrast, in monodirectional control systems in some tissues such as hepatocytes and many endocrine cells, the two classes of receptors appear not to interact with each other but to function independently or cooperatively. In hepatocytes, for instance, the inositol phospholipid turnover that is induced by aadrenergic stimulators is not blocked by P-adrenergic stimulators nor by dibutyryl cyclic AMP (184). Both a- and p-stimulators are well known to cause glycogenolysis in the liver. In some tissues such as pineal gland, the cellular responses to P-adrenergic stimulators are markedly potentiated by a-adrenergic stimulators that induce inositol phospholipid turnover and Ca2 mobilization (185).Presumably, in such tissues protein kinase C potentiates the adenylate cyclase system or acts cooperatively with protein kinase A to induce full cellular responses. Considerable variations arise in receptor interaction, but further exploration of the biochemical basis of such interaction might be of great importance for understanding the basal mechanism of signal transduction. Arachidonic acid is shown to be derived from inositol phospholipids through two consecutive reactions catalyzed by phospholipase C followed by diacylglycer01 lipase (186). Inositol phospholipids in mammalian tissues contain mostly arachidonic acid at the position 2 (21). This fatty acid, however, may also be released from phosphatidylethanolamine as well as from phosphatidylcholine. Perhaps, when the receptor is stimulated, both phospholipases C and A, act in concert. +

+

+

183

5. PROTEIN KINASE C

Although Ca2 causes direct activation of guanylate cyclase in some tissues +

(187),it seems likely that arachidonic acid peroxide and prostaglandin endoperoxide serve as activators for this enzyme (188, 189). Cyclic GMP may have a

function to act as a “negative” rather than a “positive” intracellular mediator, providing an immediate feedback control that prevents overresponse. It is known that sodium nitroprusside, which induces a marked elevation of cyclic GMP levels, is a powerful inhibitor of platelet activation (190). Indeed, analogous to cyclic AMP, 8-bromo cyclic GMP as well as sodium nitroprusside is shown to inhibit the signal-induced inositol phospholipid breakdown, and thereby counteracts the activation of protein kinase C (191). Although some functions of cyclic GMP and cyclic GMP-dependent protein kinase have been suggested for nervous tissues (168), crucial information on the role of this cyclic nucledotide is limited.

XI.

Conclusion

This chapter summarizes our knowledge of protein kinase C, which is expanding very rapidly. The evidence available strongly suggests its crucial role in signal transduction for the activation of many cellular functions and proliferation, particularly at the early phase of responses. Perhaps, the signal-induced breakdown of inositol phospholipids initiates a cascade of events starting with Ca2+ mobilization and protein kinase C activation and ending with alterations of a variety of cellular processes including gene expression in long-term. However, it seems early to discuss the precise relationship between the roles of Ca2+ and protein kinase C. Several functions of this enzyme seem plausible, such as synergistic roles with Ca2 , Ca2 -sensitivity modulation, plasticity, down regulation, desensitization, or dual functions. Obviously, Ca2 and protein kinase C each appear to play diverse roles in controlling cellular processes, and it is hoped that further exploration of the roles of this unique protein kinase may provide clues to the biochemical bases of signal transduction and cellular responses. +

+

+

ACKNOWLEDGMENTS The authors are grateful to Mrs. M . Furuta, Miss S. Fukase, and Mrs. S. Nishiyarna for their skillful assistance for preparation of this manuscript.

REFERENCES 1 . Nishizuka, Y. (1983). Philos. Trans. R . Soc. London, Ser. B 302, 101-112. 2. Nishizuka, Y. (1983). TrendsBiochem. Sci. 8, 13-16. 3 . Nishizuka, Y . (1984). Nature (London) 308,693-698. 4. Nishizuka, Y.(1984). Science 225, 1365- 1370.

184


5. Nishizuka, Y., Takai, Y., Kishimoto, A,, Kikkawa, U.,and Kaibuchi, K. (1984). Recent Prog. Horm. Res. 40, 301-345. 6. Hirasawa, K., and Nishizuka, Y . (1985). Annu. Rev. Pharmacol. Toxicol. 25, 147-170. 7. Inoue, M., Kishimoto, A,, Takai, Y.,and Nishizuka, Y. (1977). JBC 252,7610-7616. 8. Takai, Y.,Kishimoto, A,, Kikkawa, U., Mori, T., and Nishizuka, Y. (1979). BBRC91, 12181224. 9. Kishimoto, A,, Takai, Y., Mori, T., Kikkawa, U.,and Nishizuka, Y. (1980). JBC 255,22732276. 10. Kuo, J. F., Andersson, R. G. G., Wise, B. C., Mackerlova, L., Salomonsson, I., Brackett, N. L., Katoh, N., Shoji, M., and Wrenn, R. W. (1980). PNAS 77,7039-7043. 11. Minakuchi, R., Takai, Y., Yu,B., and Nishizuka, Y. (1981). J. Eiochem. (Tokyo) 89, 16511654. 12. Kikkawa, U., Takai, Y., Minakuchi, R., Inohara, S . , and Nishizuka, Y. (1982). JBC 257, 13341- 13348. 13. Wise, B. C., Raynor, R. L., and Kuo, J. F. (1982). JBC257. 8481-8488. 14. Schatzman, R. C., Raynor, R. L., Fritz, R. B., and Kuo, J. F. (1983). BJ209,435-443. 15. Parker, P. J., Stabel, S., and Waterfield, M. D. (1984). E M B O J . 3, 953-959. 16. Uchida, T., and Filburn, C. R. (1984). JEC 259, 12311-12314. 17. Inagaki, M., Watanabe, M., and Hidaka, H. (1985). JBC 260,2922-2925. 18. Kitano, T., Go, M., Kikkawa, U., and Nishizuka, Y.(1986). “Methods in Enzymology” 124, 349-252 19. Wise, B. C., Glass, D. B., Chou, C. H. J., Raynor, R. L., Katoh, N., Schatzman, R. C., Turner, R. S., Kibler, R. F., and Kuo, J. F. (1982). JBC 257,8489-8495. 20. Kaibuchi, K., Takai, Y., and Nishizuka, Y. (1981). JBC 256, 7146-7149. 21. Holub, B. J., Kuksis, A,, and Thompson, W. (1970). J. Lipid Res. 11, 558-564. 22. Kajikawa, N., Kikkawa, U . , Itoh, K., and Nishizuka, Y. (1986). “Methods in Enzymology” (in press). 23. Lapetina, E. G., Reep, B., Ganong, B. R., and Bell, R. M. (1985). JBC 260, 1358-1361. 24. Davis, R. J., Ganong, B. R., Bell, R. M., and Czech, M. P. (1985). JBC 260, 1562-1566. 25. Rando, R. R., and Young, N. (1984). BERC 122,818-823. 26. Kishimoto, A,, Kajikawa, N., Shiota, M., and Nishizuka, Y. (1983). JBC258, 1156-1164. 27. Tapley, P. M., and Murray, A. W. (1984). BBRC 118, 835-841. 28. Michell, R. H. (1975). BBA 415, 81-147. 29. Hawthorne, J. N.,and White, D. A. (1975). Vitam. Horm. (N.Y.)33, 529-573. 30. Benidge, M. J. Mol. Cell. Endocrinol. (1981). 24, 115-140. 31. Michell, R. H., Kirk, C. J., Jones, L. M., Downes, D. P., and Creba, J. A. (1981). Philos. Trans. R. SOC.London, Ser. E 296, 123-137. 32. Hokin, M. R., and Hokin, L. E. (1953). JBC 203, 967-977. 33. Abdel-Latif, A. A., Akhtar, R. A., and Hawthorne, J. N. (1977). BJ 162, 61-73. 34. Downes, P., and Michell, R. H. (1982). CeN Calcium 3, 467-502. 35. Agranoff, B. W., Murthy, P., and Seguin, E. B. (1983). JBC 258, 2076-2078. 36. Berridge, M. J., Dawson, R. M. C., Downes, C. P., Heslop, J. P., and Irvine, R. F. (1983). EJ 212, 473-482. 37. Fisher, S. K., van Rooijen, L. A. A., and Agranoff, B. W. (1984). Trends Biochem. Sci. 9, 53-56. 38. Berridge, M. J., and Irvine, R. F. (1984). Nature (London) 312, 315-321. 39. Streb, H., Irvine, R. F., Berridge, M. J., and Schulz, I. (1983). Nature (London) 306,67-69. 40. Boyer, J. L., Garcia, A,, Posadas, C., and Garcia-Sainz, J. A. (1984). JBC 259, 8076-8079. 41. Haslam, R. J., and Davidson, M. M. L. (1984). FEES Leu. 174, 90-95. 42. Haslam, R. J., and Davidson, M. M. L. (1984). J. Recepr. Res. 4, 605-629.

5. PROTEIN KINASE C

185

43. Nakamura, T . , and Ui, M. (1984). FEBS Lett. 173, 414-418. 44. kdwahard, Y., Takai, Y., Minakuchi, R., Sano, K., and Nishizuka, Y. (1980). BBRC 97, 309-3 17. 45. Sano, K., Takai, Y., Yamanishi, J., and Nishizuka, Y. (1983). JBC 258, 2010-2013. 46. Lyons, R. M., Stanford, N., and Majerus. P. W. (1975). J. Clin. Invest. 56, 924-936. 47. Haslam, R. J., and Lynham, J. A. (1977). BBRC 77, 714-722. 48. Daniel, J. L., Holmsen, H., and Adelstein, R. S. (1977).Thromb. Haemostasis 38,984-989. 49. Rittenhouse-Simmons, S . (1979). J. Clin. Invest. 63, 580-587. 50. Bell, R. L., and Majerus, P. W. (1980). JBC 255, 1790-1792. 51. Kaibuchi, K., Sano, K., Hoshijima, M., Takai, Y., and Nishizuka, Y. (1982).Cell Calcium 3, 323-335. 52. Kaibuchi, K., Takai, Y., Sawamura, M., Hoshijima, M., Fujikura, T., and Nishizuka, Y. (1983). JBC 258, 6701-6704. 53. Van Duuren, B. L. (1969). f r o g . Exp. Tumor Res. 11, 31-68. 54. Hecker, E. (1971). Methods Cancer Res. 6, 439-484. 5 5 . Boutwell, R. K. (1974). CRC Crir. Rev. Toxicol. 2, 419-443. 56. Weinstein, I. B., Lee, L. S., Mufson, A,, and Yamasaki, H. (1979).J. Suppramol. Strucr. 12, 195-208. 57. Blumberg, P. M. (1980). CRC Crit. Rev. Toxicol. 8, 153-197. 58. Castagna, M., Takai, Y., Kaibuchi, K., Sano, K., Kikkawa, U . , and Nishizuka, J. (1982). JBC 257, 7847-785 1. 59. Yamanishi, J., Takai, Y., Kaibuchi, K., Sano, K., Castagna, M., and Nishizuka, Y. (1983). BBRC 112, 778-786. 60. Kikkawa, U . , Takai, Y., Tanaka, Y., Miyake, R., and Nishizuka, Y. (1983). JBC 258, 11442-1 1445. 61. Blumberg, P. M., Jaken, S., Koning, B., Sharkey, N. A,, Leach, K. L., Jeng, A. Y., and Yeh, E. (1984). Biochem. Pharmacol. 33, 933-940. 62. Ashendel, C. L., Staller, J. M., and Boutwell, R. K. (1983). BBRC 111, 340-345. 63. Sando, J. J., and Young, M. C. (1983). PNAS 80, 2642-2646. 64. Horowitz, A. D., Greenebaum, E., and Weinstein, I. B. (1981). PNAS 78, 2315-2319. 65. Solanki, V., and Slaga, T. J. (1981). PNAS 78, 2549-2553. 66. Niedel, 3. E., Kuhn, L. J., and Vandenbark, G. R. (1980). PNAS 80, 36-40. 67. Leach, K. L., James, M. L., and Blumberg, P. M. (1983). PNAS 80, 4208-4212. 68. Kraft, A. S., Anderson, W. B., Cooper, H. L., and Sando, J. J. (1982). JBC 257, 1319313196. 69. Shoyab, M., Todaro, G. J., and Tallman, J. F. (1982). Cancer Lett. 16, 171-177. 70. Miyake, R.,Tanaka, Y., Tsuda, T., Kaibuchi, K., Kikkawa, U., and Nishizuka, Y. (1984). BBRC 121, 649-656. 71. Fujiki, H., Tanaka, Y., Miyake, R., Kikkawa, U . , Nishizuka, Y., and Sugimura, T . (1984). BBRC 120, 339-343. 72. de Chaffoy de Courcelles, D., Roevans, P., and Van Belle, H. (1985). JBC 260, 1576215770. 73. Mori, T., Takai, Y., Minakuchi, R., Yu, B., and Nishizuka, Y. (1980).JBC 255,8378-8380. 74. Hidaka, H.,Sasaki, Y., Tanaka, T., Endo, T., Ohno, S., Fujii, Y., and Nagata, T. (1981). PNAS 78, 4354-4357. 75. Schatzman, R. C., Wise, B. C., and Kuo, J. F. (1981). BBRC 98, 669-676. 76. Wise, B. C., Glass, D. B., Chou, C. H. J., Raynor, R. L., Katoh, N., Schatzman, R. C., Turner, R. S . , Kibler, R. F., and Kuo, J. F. (1982). JBC 257, 8489-8495. 77. Katoh, N., Raynor, R. L., Wise, B. C., Schatzman, R. C., Turner, R. S., Helfman, D. M., Fain, J. N., and Kuo, J. F. (1982). BJ 202, 217-224.

186


78. Hidaka, H., Inagaki, M., Kawamoto, S., and Sasaki, Y. (1984). Biochemistry 23,5036-5041. 79. Kuo, J. F., Raynor, R. L.. Mazzei, G . J., Schatzman, R. C., Turner, R. S., and Kem, W. R. (1983). FEBS Lett. 153, 183-186. 80. Swilem, A. M. F., Hawthorne, J. N., and Azila, N. (1983). Biochem. fharmucol. 32, 38733874. 81. Starke, K. (1980). Trends fharmacol. Sci. 1, 268-271. 82. Knight, D. E., Hallam, T. J., and Scrutton, M. C. (1982). Nature (LondonJ 296, 256-257. 83. Hokfelt, T., Johansson, 0..Ljingdahl, A., Lundberg, J. M., and Schultzberg, M. (1980). Nature (London) 284, 515-521. 84. Yoshida, H., and Quastel, J. H. (1962). BBA 57, 67-72. 85. Birnberger, A. C., Birnberger, K. L., Eliasson, S. G., and Simpson, P. C. (1971). J. Neurochem. 18, 1291-1298. 86. Schacht, J., and Agranoff, B. W. (1972). J. Neurochem. 19, 1417-1421. 87. Hawthorne, J. N., and Pichard, M. R. (1979). J. Neurochem. 32, 5-14. 88. Knight, D. E., Niggli, V., and Scrutton, C. (1984). EJB 143, 437-446. 89. Kajikawa, N., Kaibuchi, K., Matsubara, T., Kikkawa, U.,Takai, Y., Nishizuka, Y., Itoh, K., and Tomioka, C. (1983). BBRC 116, 743-750. 90. Katakami, Y., Kaibuchi, K., Sawamura, M., Takai, Y., and Nishizuka, Y. (1984). BBRC 121, 573-578. 91. O’Flaherty, J. T., Schmitt, J. D., McCall, C. E., and Wykle, R. L. (1984). BBRC 123,64-70. 92. White, J. R., Huang, C. K., Hill, 3. M., Jr., Naccache, P. H.,Becker, E. L., and Sha’afi, R. I. (1984). JBC 259, 8605-861 1. 93. Knight, D. E., and Baker, P. E. (1983). FEBS Lett. 160, 98-100. 94. Kojima, I., Lippes, H., Kojima, K., and Resmussen, H. (1983). BBRC 116, 555-562. 95. Zawalich, W., Brown, C., and Rasmussen, H. (1983). BBRC 117, 448-455. 96. Hubinont, C. J., Best, L., Sener, A., and Malaise, W. 1. (1984). FEBS Lett. 170, 247-253. 97. Tamagawa, T., Niki, H., and Niki, A. (1985). FEBS Leu. 183, 430-432. 98. de Pont J. J. H. H. M., and Flueren-Jacobs, A. M. M. (1984). FEBS Lett. 170, 64-68. 99. Hirota, K., Hirota, T., Aguilera, G., and Catt, K. I. (1985). JBC 260, 3243-3246. 100. Martin, T. F. J., and Kowalchyk, J. A. (1984). Endocrinology (Baltimore) 115, 1527-1536. 101. Tanaka, C., Taniyama, K., and Kusunoki, M. (1984). FEBS Leu. 170, 64-68. 102. Culty, M., Vilgrain, I., and Chambaz, E. M. (1984). BBRC 121, 499-506. 103. Fujita, I., Irita, K., Takeshige, K., and Minakami, S. (1984). BBRC 120, 318-324. 104. Robinson, J. M., Badway, J. A., Kamovsky, M. L., and Karnovsky, M. J. (1984). BBRC 122, 734-739. 105. Dale, M. Maureen, and Penfield, A. (1984). FEBS Leit. 175, 170-172. 106. Rasmussen, H., Forder, J., Kojima, I., and Scriabine, A. (1984). BBRC 122, 776-784. 107. Baraban, J. M., Could, R. J., Peroutka, G. J., and Snyder, S. H. (1985). PNAS 82,604-607. 108. Garrison, J. C., Johnsen, D. E., and Campanile, C. P. (1984). JBC 259, 3283-3292. 109. Fain, J. N., Li, S. Y., Litosch, I., and Wallace, M. (1984). BBRC 119, 88-94. 110. Rink, T. J., Sanchez, A., and Hallan, T. J. (1983). Nature (London) 305, 317-319. 111. Di Virgilio, F., Lew, D. P., and Pozzan, T. (1984). Nature (London) 310, 691-693. 112. Kaibuchi, K., Takai, Y., and Nishizuka, Y. (1985). JBC 260, 1366-1369. 113. Truneh, A., Albert, F., Golstein, P., and Schmitt-Verhulst, A. M. (1985). Nature (London) 313, 318-320. 114. Rozengurt, E., Rodriguez-Pena, A., Coombs, M., and Sinnett-Smith, J. (1984). PNAS 81, 5748-5752. 115. Mastro, A. M., and Mueller, G. C. (1974). Exp. Cell Res. 88, 40-46. 1 16. Kriuijer, W., Cooper, J . A., Hunter, T., and Verma, 1. M. (1984). Nature (London) 312,71 I 716.

5. PROTEIN KINASE C

187

117. Cochet, C., Gill. G. N., Meisenhelder, J.. Cooper, J. A., and Hunter, T. (1984). JBC 259, 2553-2558. 118. McCaffrey, P. G., Friedman, B., and Rosner, M. R. (1984). JBC 259, 12502-12507. 119. Iwashita, S.. and Fox, C. F. (1984). JBC 259, 2559-2567. 120. Brown, K. D., Blay, J., Irvine, R. F., Heslop, J. P.,and Berridge, M. J. (1984). BBRC 123, 377-384. 121. Moon, S . 0.. Palfrey, H. C., and King, A. C. (1984). PNAS 81, 2298-2302. 122. Friedman, B., Frackelton, A. R., Ross, A. H.,Connors, J. M., Fujiki, H., Sugimura, T., and Rosner, M. R. (1984). PNAS 81, 3034-3038. 123. Hunter, T., Ling, N., and Cooper, J. A. (1984). Nature (London) 311, 480-483. 124. Ullrich, A., Coussens, L., Hayflick, J. S . , Dull, T. J., Gray, A., Tam, A. W., Lee, J., Yarden, Y., Libermann, T. A , , Schlessinger, J., Downword, J., Mayes, E. L. V., Whittle, N., Waterfield, M. D., and Seeburg, P. H. (1984). Narure (London) 309, 418-430. 125. Lee, L. S . , and Weinstein, I. B. (1978). Science 202, 313-315. 126. Shoyab, M., DeLarco, J. E., and Todaro, G. J . (1979). Nature (London) 279, 387-391. 127. Krupp, M. N., Connoly, D. T., Lane, M. D. (1982). JBC 257, 11489-1 1496. 128. King, A. C., and Cuatrecasas, P. (1982). JBC 257, 3053-3060. 129. Jacobs, S . , Sahyoun, N. E., Saltiel, A. R., and Cuatrecasas, P. (1983).PNAS 80,621 1-6213. 130. May, W. S . , Jacobs, S., and Cuatrecasas, P. (1984). PNAS 81, 2016-2020. 131. Klausner, R. D., Harford, J., and van Renswoude, J. (1984). PNAS 81, 3005-3009. 132. Shackelford, D. A , , and Trowbridge, 1. S. (1984). JBC259, 11706-11712. 133. Corvera, S., and Garcia-Sainz, J. A. (1984). BBRC 119, 1128-1 133. 134. Nishikawa, M., Hidaka, H., and Adelstein, R. S. (1983). JBC 258, 14069-14072. 135. Inagaki, M . , Kawamoto, S., and Hidaka, H. (1984). JBC 259, 14321-14323. 136. Tsien, R. Y., Pozzan, T., and Rink, T. J. (1982). Narure (London) 295, 68-71. 137. Moolenaar, W. H., Tertoolen, L. G. J., and de Laat, S. W. (1984). Nature (London) 312, 371-374. 138. Sagi-Eisenberg, R., Lieman, H., and Pecht, 1. (1985). Nature (London) 313, 59-60. 139. Carafoli, E., and Compton, M. (1978). Curr. Top. Membr. Tramp. 10, 151-216. 140. Vincenzi, F. F., and Hinds, T. R. (1980). In “Calcium and Cell Function” (W. Y. Cheung, ed.), Vol. 1 , pp. 127-165. Academic Press, New York. 141. Limas, C. J. (1980). BBRC 96, 1378-1383. 142. Lagast, H., Pozzan, T., Waldvogel, F. A,, and Lew, P. D. (1984). J. Clin. Invesr. 73, 878883. 143. Iwasa, Y . , and Hosey, M. M. (1984). JBC 259, 534-540. 144. Movsesian, M. A., Nishikawa, M., and Adelstein, R. S. (1984). JBC 259, 8029-8032. 145. Albert, K. A , , Wu, W. C. S., Nairn, A. C., and Greengard, P. (1984). PNAS 81, 36223625. 146. Labarca, R., Janowsky, A,, Patel, J., and Paul, S . M. (1984). BBRC 123, 703-709. 147. Endo, T., Naka, M., and Hidaka, H. (1982). 105, 942-948. 148. Katoh, N., Wise, B. C., and Kuo, J. F. (1983). BJ 209, 189-195. 149. Kawamoto, S . , and Hidaka, H. (1984). BBRC 118, 736-742. 150. Werth, D. K., Niedel, J. E., and Pastan, I. (1983). JBC 258, 11423-1 1426. 151. Takai, Y., Kikkawa, U . , Kaibuchi, K., and Nishizuka, Y. (1984). Adv. Cyclic Nucleotide Protein Phosphorylation Res. 18, 119-158. 152. Fitzgerald, D. J . , Knowles, S. E., Ballard, F. J., and Murray, A. W. (1983). Cancer Res. 43, 3614-36 18. 153. Enomoto, T., Martel, N., Kanno, Y., and Yamasaki. H. (1984). J . Cell. Physiol. 121, 323333. 154. Gainer, H. St. C., and Murray, A. W. (1985). BBRC 126, 1109-1113.

188


155. Ramachandran, C., Yan, P., Bradbury, E. M., Shyamala, G., Yasuda, H., and Walsh, D. A. (1984). JBC 259, 13495-13503. 156. Hirata, F., Matsuda, K., Notsu, Y.,Hattori, T., and del Carmine, R. fNAS 81, 4717-4721. 157. Jolles, J . , Zwiers, H., Dekker, A,, Wirtz, K. W. A,, and Gispen, W. H. (1981).BJ 194, 283291. 158. Aloyo, V. J . , Zwiers, H., and Gispen, W. H. (1983). J. Neurochem. 41, 649-653. 159. Lopez-Rivas, A., Stroobant, P., Waterfield, M. D., and Rozengurt, E. (1984). EMBO J. 3, 939 -944. 160. Kishimoto, A., Mori, T., Takai, Y.,and Nishizuka, Y. (1978). J. Biochem. (Tokyo)84, 4753. 161. Ahmad, Z., Lee, F. T., DePaoli-Roach, A., and Roach, P. J . (1984). JBC 259, 8743-8747. 162. Zwiller, J . , Revel, M. 0.. and Malviya, A. N. (1985). JBC 260, 1350-1353. 163. Schatzman, R. C., Grifo, J. A., Menick, W. C., and Kuo, J. F. (1983). FEES Lett. 159, 167170. 164. Papanikolau, P., Humble, E., and Engstrom, L. (1982). FEES Lett. 143, 199-204. 165. Sulakhe, P. V., Petrali, E. H., Daris, E. R., and Thiessen, B. J. (1980). Biochemistry 19, 5363-5371. 166. Cope, F. O., Staller, J. M., Mahsen, R. A., and Boutwell, R. K. (1984).BBRC 120,593-601. 167. Le Peuch, C. J., Ballester, R., and Rosen, 0. M. (1983). fNAS 80, 6858-6862. 168. Nestler, E. J . , and Greengard, P. (1983). Nature (London) 305, 583-588. 169. Ehrlich, Y. H., and Routtenberg, A. (1974). FEES Lett. 45, 237-243. 170. Routtenberg, A. (1979). f r o g . Neurobiol. 12, 85-113. 171. Sorenson, R. G., Kleine, L. P., and Mahler, H. R. (1981). Bruin Res. Bull. 7, 56-61. 172. Kristjansson, G. I., Zwiers, H., Oestreicher, A. B., and Gispen, W. H. (1982).J. Neurochem. 39, 371-378. 173. Lovinger, D. M., Akers, R. F., Nelson, R. B., Barnes, C. A., McNaughton, B. L., and Routtenberg, A. (1985). Bruin Res. 343, 137-143. 174. Routtenberg, A., and Lovinger, D. (1985). Behav. Neural Biol. 43, 3-11. 175. Zwiers, H., Schotman, P., and Gispen, W. H. (1980). J. Neurochem. 34, 1689-1699. 176. Zwiers, H., Jolles, J., Aloyo, V. J., Oestreicher, A. B., and Gispen, W. H. (1982). f r o g . Bruin Res. 56, 405-417. 177. Routtenberg, A. (1984). In “Neurobiology of Learning and Memory” (G. Lynch, J . McGaugh, and N. Weinberger, eds.), pp. 479-490. Guilford Press, New York. 178. Akers, R. F., and Routtenberg, A,, (1985). Bruin Rex 334, 147-151. 179. DeRiemer, S. A., Strong, J. A., Albert, K. A,, Greengard, P., and Kaczmarek, L. K. (1985). Nature (London) 313, 313-316. 180. Krebs, E. G., and Beavo. J. A. (1979). Annu. Rev. Biochem. 48, 923-959. 181. Kishimoto, A., Nishiyama, K.,Nakanishi, H., Uratsuji, Y., Nomura, H., Takeyama, Y., and Nishizuka, Y. (1985). JBC 260, 12492-12499. 182. Katz, A. M., Tada, M., and Kirschberger, M. A. (1975).Adv. Cyclic NucleotideRes. 5 , 453472. 183. Heyworth, C. M., Whetton, A. D., Kinsella, A. R., and Houslay, M. D. (1984). FEES Lett. 170, 38-42. 184. Kaibuchi, K., Takai, Y., Ogawa, Y., Kimura, S., Nishizuka, Y., Nakamura, T., Tonomura, A., and Ichihara, A. (1982). BBRC 104, 105-112. 185. Sugden, D., Vanecek, J., Klein, D. C., Thomas, T. P., and Anderson, W. B. (1985). Nature (London) 314, 359-361. 186. Bell, R. L., Kennerley, D. A., Stanford, N., and Majerus, P. W. (1979). PNAS 76, 32383241.

5. PROTEIN KINASE C

189

187. Murad, F., Arnold, W. P., Mittal, C. K., and Braughler, J . M. (1979). Adv. Cyclic Nucleoride Res. 11, 175-204. 188. Hidaka, H., and Asano, T. (1977). PNAS 74, 3657-3661. 189. Graff, G . , Stephenson, J. H., Glass, D. B.,Haddox, M. K., and Goldberg, M. D. (1978).JEC 253, 7662-7676. 190. Haslam, R.J., Salarna, S. E., Fox, J . E. B., Lynham, L. A., and Davidson, M. M. L. (1980). I n “Platelets; Cellular Response Mechanisms and their Biological Significance” (A. Rotman, F. A. Meyer, C. Gitler, and A. Silderberg, eds.), pp. 213-231. Wiley, New York. 191. Takai, Y . , Kaibuchi, K., Matsubara, T., and Nishizuka, Y. (1981). EERC 101, 61-67.


Viral Oncogenes and Tyrosine Phosphorylation TONY HUNTER JONATHAN A. COOPER* Molecular Biology and Virology Laboratory The Salk Institute Sun Diego, California 92138

I. Introduction and Historical Perspective .

................

192

. . .. . . . .. . . .. . . . . .

193 193 200 20 1 202 205 201 208 21 I 214 214 215 215 216 217 218 219 219 220

.............

.............................. . . . .. . . . . . . . . . 111. Other Transformation-Related Ty A. The v-sis Oncogene, PDGF,

Receptor . . . . . . . . . . . . . . .

_.................. C. The neu Oncogene ................................. D. pp6oE-s'c and Polyoma Virus Middle T Antigen . . . . . . . . . . . . . . . . . . . . E. p56Istra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. The v-mil, v-raf, and v-mos On IV. General Properties of Protein-Tyrosi . .... ..... ... ............. A. Sequence Homologies . . . . . . . . B. Catalytic Domain . . . . . . . . . . .

*Present address: Fred Hutchinson Cancer Research Center, 1 124 Columbia Street, Seattle, Washington 98104 191 THE ENZYMES, Vol. XVII Copyright 0 1986 by Academic Press. Inc. All rights of reproduction in any form reserved.

192

TONY HUNTER AND JONATHAN A. COOPER

C. Requirements for Substrate Selection ............................. D. Regulation of Protein-Tyrosine Kinases V. Cellular Substrates for Protein-Tyrosine Kin A. Substrates for the Viral Protein-Tyrosine Kinases . . . . . . . . . . . . . . . . . . . B . Substrates for the Growth Factor Receptor

..................................... .... ............ VI. Conclusions References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.

225

229 233 235 237

Introduction and Historical Perspective

For nearly thirty years phosphorylation has been recognized as a means of reversibly modulating the function of proteins. Over this period many protein kinases specific for serine and threonine residues in their substrates have been identified and characterized. However, not until the 1980s have enzymes with specificity for tyrosine, the third hydroxyamino acid, been reported. Tyrosine phosphorylating activity was originally detected in partially purified preparations of two viral transforming proteins (I, 2). Since that time it has become clear that there is a large family of protein-tyrosine kinases. Eight such cellular enzymes were first recognized in altered guises as parts of the transforming proteins of a series of acutely oncogenic retroviruses, while five other protein-tyrosinekinases are growth factor receptors. Over fifteen distinct protein-tyrosine kinase genes have been identified with the prospect of several more to follow. The seminal discovery of a protein kinase activity associated with pp6OV-"" (3), the product of the Rous sarcoma virus (RSV) src gene, immediately suggested that transformation could be due to aberrant protein phosphorylation events that would modulate the functions of critical cellular proteins in an abnormal fashion. This idea was reinforced by the unexpected finding that pp60v-src phosphorylated tyrosine rather than serine or threonine ( 4 ) . The subsequent demonstration that several other distinct retroviral oncogene products had similar tyrosine phosphorylating activities lent further credence to this notion. Out of twenty known viral oncogenes, there are eight whose products are proteintyrosine kinases: namely, the v-src, v-yes, v-fsr, v-fps and v-fes, v-abl, v-ros, v-erb-B and v-jhs oncogenes. Abnormal patterns of typrosine phosphorylation are manifest in cells transformed by most of the relevant viruses, and several substrates for the viral protein-tyrosine kinases have been identified. Nevertheless, despite extensive efforts, the proof that tyrosine phosphorylation is in fact critical for transformation by these viruses is not at hand. The problem has been to pinpoint substrates for the viral protein-tyrosine kinases whose functions are changed in ways that would explain the transformed phenotype. The aim of this chapter is to summarize our knowledge of the viral proteintyrosine kinases and their roles in viral transformation. For this purpose we have

6.

VIRAL ONCOGENES AND TYROSINE PHOSPHORYLATION

193

reviewed the properties of the individual viral protein-tyrosine kinases and contrasted them to the cognate cellular enzymes that are encoded by the cellular genes homologous to the viral oncogenes. We have included a discussion of a number of transformed cell types where altered tyrosine phosphorylation is evident, but which do not involve one of the aforementioned viral oncogenes. The general properties of protein-tyrosine kinases and their common structural features are described. We conclude with a progress report on the identification of substrates for the viral protein-tyrosine kinases. Throughout we have tried to address the question of how the viral protein-tyrosine kinases differ from their cellular counterparts, enzymes that clearly coexist peaceably with normal cells. In many places we have found it pertinent to compare and contrast the properties of the viral enzymes to those of the growth factor receptor protein-tyrosine kinases, particularly because of the abnormal growth state of transformed cells. For a more detailed review of the growth factor receptor protein-tyrosine kinases, however, the reader is referred to Chapter 7 by Morris White and C. Ronald Kahn.

II. Individual Viral Protein-Tyrosine Kinases and Their Cellular Homologues

A.

pp60v-srcAND pp60c-src

The complete sequence of the src gene of RSV predicts a primary translation product of 526 amino acids (5-7). In fact pp6OV-"'" is modified by the removal of the initiating methionine and the subsequent linkage of a myristyl group to the glycine which is exposed (8, 9), leaving a mature protein of 525 amino acids. The pp6OV-"" was originally identified by immunoprecipitation from RSV-transformed cells with serum from an RSV tumor-bearing rabbit (TBR serum) (10). The protein kinase activity of pp60v-"rcwas also first detected in immunoprecipitates made with TBR serum, in this case leading to the phosphorylation of the immunoglobulin heavy chain (3). TBR sera contain variable amounts of antibodies to viral structural proteins in addition to those against pp60v-"rc. Some TBR sera cross-react with pp60v-src from many strains of RSV as well as pp60c-"rc, while others are specific for pp60"-src. Interestingly while immunoprecipitates of pp60v-src made with TBR serum do not exhibit autophosphorylation or phosphorylation of exogenous substrates, similar immunoprecipitates of pp60c-srcwill both autophosphorylateand phosphorylate added proteins in addition to phosphorylating heavy chain. A number of other reagents are available for immunoprecipitation of pp60v-src, some of which recognize pp60c-srcas well. Antibodies against bacterially expressed pp60v-srccross-react with pp60c-srcand recognize a population of pp60v-srcthat is largely perinuclear,

194


which is not detected with antitumor sera (11, 12). Several hybridomas derived from mice immunized with bacterially expressed pp6OV-"" produce monoclonal antibodies (13, 14) which have specificities similar to polyclonal antisera against this protein. The major antigenic determinants for both the polyclonal and monoclonal antibacterial pp60v-src.are in the N-terminal half of pp6WrC. Both types of antibody allow autophosphorylation of pp6OSrCand phosphorylation of exogenous substrates. Antipeptide antisera directed against amino acids 409-4 19 (15, 16), 498-512 (17) and against the C-terminal 6 residues (521-526) (18) immunoprecipitate pp60v-src.The latter does not recognize pp60C-src due to its different C-terminus. Antibodies to the 409-419 (16) and 498-512 (17) src peptides inhibit pp60v-srcprotein kinase activity, while immunoprecipitates of pp60v-srcmade with antibodies to the 521-526 src allow phosphorylation of exogenous substrates (18). A number of antibodies specific for synthetic peptides corresponding to sequences from pp60v-SrChave been affinity purified with TBR sera (19). Although pp60v-srcis synthesized on soluble ribosomes, the bulk of the protein is found associated with cellular membranes, which include the plasma membrane and perinuclear membranes (12, 13,19-23). However, pp6OV-"" is not an integral membrane protein but is bound in a peripheral fashion. Newly synthesized molecules of pp60v-srcare associated in a complex with two cellular proteins, pp89 and pp50 (24, 25); pp89 has been identified as one of the major stressinduced proteins, but the function of pp50 is unknown. It has been proposed that this complex serves to transport pp60v-srcto the membrane. A mutant pp60v-srCin which the N-terminal glycine has been replaced by an alanine is not myristylated and is found to be largely soluble in the cell (26).Such pp6OV-"" mutants appear to be fully active as protein-tyrosine kinases but are nontransforming (26, 27). This implies that association of pp60v-srcwith the membrane, presumably via the myristyl group, is essential for transformation. The pp60v-srCaccounts for about 0.05% of total cell protein in cells transformed by RSV (28). The half-life of pp60v-src,depending on the strain of RSV, ranges from 2 to 7 h (28, 29). pp60v-srChas two functional domains. The C-terminal30,OOOdaltons (residues 270-526, numbering from Met 1) have been shown to contain the entire catalytic domain on the basis of limited proteolysis (30,31).The sequence in this region of pp60v-Srcshows striking homology to that of the catalytic subunit of the CAMPdependent protein kinase (32) (see Fig. 1). The homology includes a number of key sequence motifs, known to be critical for function, which will be discussed in Section IV. A similar catalytic domain is also recognizable in all the other proteintyrosine kinases. As previously mentioned the very N-terminus of pp6OV-"" plays an essential role in the association of pp60v-srcwith cellular membranes. Two lines of evidence suggest that the rest of the N-terminal domain is involved in regulating catalytic activity. First, some of the point mutations that render pp60v-SrCtemperature-sensitive for phosphotransferaseactivity map between residues 2 and 270

6. VIRAL ONCOGENES AND TYROSINE PHOSPHORYLATION

195

(33). In addition deletion mutations in this region can either render catalytic activity temperature-sensitive [e.g., deletion of residues 173-227 or 169-225 (34, 35)],affect the phenotype of the transformed cell [e.g., deletion of residues 16-82, 16-138 (36),15-169 (37), or 135-236 (38)],or even abolish transforming activity altogether [e.g., deletion of residues 82-169 (331. Second, this domain of pp6OV-"" also contains several phosphorylation sites, occupation of which can alter phosphotransferase activity. A great deal of effort has gone in to determining both the sites and effects of FIG. 1. (See pp. 196-197.) The amino acid sequences of the catalytic domains of 4 protein-serine kinases, myosin light chain kinase (MLCK) (250).phosphorylase y-subunit (PhK-y) (249),cGMPdependent protein kinase (cGPK) (248), and CAMP-dependent protein kinase catalytic subunit (cAPK) (247)are aligned with the predicted sequences of 12 protein-tyrosine kinases: pp60v-Src (src) (Schmidt-Ruppin A strain of RSV) (6). chicken pp60C-Src (c-src) (52). P!90nag-yes (yes) (67), P70nQg-lnr (fgr) (75), P120nag-ab/ (ubo (110), P1408aR-fps (fps) (80), PI IOn'R-fes (fes) (GardnerArnstein strain) (81), P68gag-ros (ros) (136), the human insulin receptor (1NS.R) (245), gp66/68v-erb-s (erb-B) (AEV-H strain) (144), human EGF receptor (EGF.R) (159), gP180Raggms (fms) (168), and the human c-fms protein (c-fms) (only the C-terminal97 residues are shown) (183). Finally the sequences of 4 other protein kinase-related proteins are given: P1O@'R-mi/ (mil) (232), P9@aR-raf(ruf)(the sequence shown starts at position 387; there are gag gene sequences upstream of ~ " ~(235), and its murine cellular counterpart (c-mos) (235). The this residue) (233), ~ 3 7 (mos) sequences were aligned by eye for maximum homology. The single letter amino acid code has been used, with . representing a gap introduced to optimize homology and with - representing an identity between the v-onc sequence and the corresponding c-onc sequence (remember that while the chicken c-src and mouse c-mos sequences are strictly comparable to the chicken-derived v-src and mousederived v-mos genes respectively, the human EGF receptor is compared to the chicken-derived v-erb-B gene, and the human c-fms protein is compared to the cat-derived v-fms gene). Residue numbers are indicated: x/, starting number; /x, finishing number; (x) terminus. In the case of the MLCK and c-fms sequences, the precise position of the sequence shown in the protein is not known, but the C-terminal 368 residues of the MLCK sequence are presented (250). Residues that are common to all 4 protein-serine kinases are given underneath the cAPK lines in capital letters; residues that are found in 3 of the enzymes are shown in lower case letters, while residues that are highly conserved with regard to properties are indicated *. The same analysis for the protein-tyrosine kinase group is depicted above the src lines. No analysis was performed for the mil, ruf, mos, and c-mos sequences, but residues that are conserved between all 20 sequences are indicated by vertical lines connecting the conserved residues between the protein-serine kinase and protein-tyrosine kinase groups. The lysine (K) which is modified by the ATP analog FSBA is in bold type, as are the threonine (T) in the CAMP-dependent protein kinase catalytic subunit which is autophosphorylated and the tyrosine (Y) which is autophosphorylated in the viral protein-tyrosine kinases. The C-termini of pp6Ov-Src, BOnaR-YeS, P'lwuR-/gr, P68saR-ros, and gP180n'R~ms are in bold type where they diverge from the corresponding cellular sequence (these cellular sequences are not known in the case of the c-yes and c-fgr proteins). Within the protein-tyrosine kinase family it should be noted that the v-fps and v-fes genes correspond to the same genetic locus in chickens and cats respectively (82). This is also true for the v-mil and v-ruf genes which correspond to the same locus in chickens and mice respectively (232-234). For those interested in comparing additional sequences of protein kinase catalytic domains, the human and chicken c-mos gene sequences have been completed, while the neu gene (209)and p561Sua(228.229)sequences are also available. In addition there are at least two other ruf-related genes, and a cDNA corresponding to one of these has been sequenced (pks).

.-.,- .-. ~~

%

dl Laf rn~e

I

W

m

c--

KLCK PhK-y CGPK cAPK

~

_

~

.... ....ASLEEAQIHKLLREDKLVQLYAW.PE ....VAVKCLRPGTILSPK.. .VAVKTLFEDTI(EVE.. ....EFLFEAAVIIFEIKEPNLVQLffiVCTRe VAVRSCRB.TLPWLR...KFLQEARILRPCNBPNIVRLIGVCTPR ...... .....VAVKSCRE.TLPPDIR.. .KFLQEAKILKQYSEPNIVRLIGVCMK

.

3 6 3 / L V Q E w E V ~ L c E L L T M C T T n R W m C L A R D I \ W e r K . . 3 2 3 / E S T V A D G L I T T L ~ P A P R R N K D T I L Y G V S P N Y D R W E . . 880/LPTIPLLIDELLQSaRPITRKSGIVLTRAVLRD~LNBEDVLLGERIGIIGNPGEVFSG~~NTP 656/PASIPLLVDELLRSOOPLTKKSGIVL~VPRDI(WVLNBEDLVLGEOIGRGNPGEVFSGRLRMNTL.

......... VAVKILRVMPTPeQPP... APRNEVAVLRmRWNI..LLRY;YUT .......VAIKOVNKCTEDLRASQR. .VAVKILRVMPTPeGLQ.. .AWNEVAVLRl?FRBVNI. .LLRY;YMT ...... ---------R_______________ 22/-------------------------_--------------------------------------........__________________ ...SEWAELNIU;L.REDNIVRWMSTAS MIETPBEIVLPllEYIEGGELZERIV~DYEL................ ...................................................... .............................................. TED KS eE~ T~~ ~ AI ~~~ LA~ E PVI AI C~Y LL~ DTYETNTFPFLVPDLnKKGELFDYLTEgVTLS........................ R T P K D S K Y L Y n L ~ A C f f i G E L W T I L R D R G S P e . . . . . . . . . . . . .......................................................... P S F K D N S N L Y H V ~ ~ P G G E n F S E ~ I G R F S . . . . . . . . ............................................................. . E ~ ~ M Q I V L T Z E Y L me *Elf 1 r 1

515/P~PVPAQReRAPGTNTQEKNKIRPRGQRDSSYYWBIEASEVLLSTRIGSGSFGTW~GED

~-387/TQEKNRIRPRGQRDSSYYWI(WeASEVl'lLSTRIGSGSPGTWKGRWED..

53/RSrSlPLVAPRR1GRLFIGTTPPRAPGLPRRLAWFSIDWBOVCLnBRLGSGGPGSWRTYEGVP. ~ . . ~ ~ ~~

I

BET

c-m

YSE

w

abl f"E

Z E

LQE

1NS.R

eE&B EGF.R

fm

mil La€ mQB

c-m

1 W Y G 11'1' ...... YHSKGSLLDFLKGEK;RYLR.................................................. ..................~-----------------L V D ~ Q I ~ I I A ~ .................... ..... .................................................................... ----EPIYIVTE......PMTKGSLLDFLKEGEGKFLK............. ....................................................... LL WW LLVVDD~ Q~ VQ A~ EI GI A~ YYnI E P I Y I VTE...... PHCBGSLIgFLKWBGQDLT .................................................................... PP~IITE......P~GNLLDYLRECNRPEVS....................................................................AWLLYIIA~ISS~YL OPIYTV ..................................................................... _ _ _ _ _ ~(...... .... LV0GGDPLSFLRSM:PRLR -~ Q P I Y I VIIE...... LVQGGDFLTFLRTeGARLR...................... ............................................... LY. . . . ILRKnnL IT PD PL L~ IL CQL~DGI C~ ~~ ~Y LL E W Y L I LE...... ~ G G D L L S Y L I I G A R R P K P P S P L L T.............................................................. . .. .......................................................... LPEnIQ~IMGIIAYL P ~ ~L. . . . LllABGDLKSYLRSLRPBMNNPGRPPPT.. TvQ.LIM......LnPYGCLLDYIREERDHIG...................................................... ...............S---------------_-a Y L L ~ ~ Q I A ~ ~ L ---F------V---------- .----...... ..................................................................... TEGGPVLVITe...YCCYGDLLNFLRRPAeAHPGPSLSVGQDPEAGAGYI(NIBLEKWRRDSGFSSPGVDTYVE~PVSTSSSWSFSBEDU;KEDGRPLELRDLLEFSSQVAPG~FL W N W I V T Q ......W C g G S S L Y K I I L B V Q E T K P P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............H ~ L I D I ~ T ~ ~ Y L WNLAIVTP. .....W ~ G S S L Y R B L W Q E T K 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H ~ L I D I ~ T ~ G ~ Y L ........................................ TRTPEDSNSLGTII.IIBPGGNV.TLBPVTYDATRSPEPLSCRKQLS................ GKCLRYSL~LLPL ______________ _______ -------G----------E---........................................................ L-----------------EPIYIVIE -T-.

____. ~~

~~

FIG. I .

I

L

~

ULCK PhR- r cGPK cAPK

I

C -6LC

yps

for abl

z

Lps

1NS.R

m-e EGF.R

fm6

mF1

raf mi3

1

.YIQVTDFGFAKR. k DFGfa

Ill

*ERDl A N l v ERllhPNERDLRAANILVGENL

.CGT.. .VKGRlTiTL. .CGT.. k cGT

.. . - ....

I

.PE . F L S P E ” Y D . . .QISD~DIIWSLGVITYILS.GLSPFLGDDDTETLNNVLS.GN PS.YLAPeIIECS~DNEPYGREVD~~VIW~L~.GSPP~E~Q~IRIII~.GN .PE .YLAPEIILNK.. .GEDISMYWSLGILMYELLT.GSPPFSGPDPUKTYNIILR.GI .YLAFEIILSK. .GYNKAVU?WALGVLIYBUAA.GYPPFFALIQPIQIYEKIVS.G. Pe y L a P E i i 9 D WslGv* Y 11 G p P F * G

I

I

I

I

R* DFG I ,G P kW A p e G SDvWsfG*ll E g Py VCKVMPGLARL. ...I E D N E Y T A R a G A K F . P I R W T A W M L Y G . . . . . . R P T I R S D V W S f f i I L L T E L T T R G R V P Y P G ~ ~ V ~ V E R . G Y ---------------------...... ___---------....__________----------------......---___--_____--------------------------...... .-ER~YIERDL~NILVGDNL......VCKIADffiLARL....IEDNEYTARaGAKF.PIKWTAPeAALYG......RPfIKSDVWSffiILLTELVTKGRVPYPG~NV~R~R.GY ERUDYIERDLPAANILVGERL. ....VCKIADffiIARL.. ..IEDNEYNPRQGARF.PIRWTAWAALffi.. .... R P r I K S m r W S F G I L L T E L I S ~ R V P Y ~ ~ N R E V ~ ~ ~ B . G Y ERKNFI ERDLMRNCLVGE ....LVKVADffiLSRL. ...~DTYTABACARF.PIRWTAPESLAYN......KFSIKSDVWAffiVLUIBIATYGlLSPYPGIDL~~LLEK.DY ESRECIERDLAARNCLVTEKN. ....TLKISD EEDCWAS’FGGWI(PIPVRWTAPeALNYG......WYSSESDVWSffiILLWBAPSLGAVPYANLS~~TREAIEQ.GV E RCC I E RDLAARNCLVTERN . EKMRFIERDLAARNCLVSEKQY NARRPVERDLAARNCUVABDF. ....TVRIGDFGKPRD.. ..IYETDYYRKGGKGLLPV~MA(AWSLR.. .... V P T T S S D ~ S f f i W U l B I T S L h B Q P Y Q G L S ~ V L R P V I I D . G G EERRLVERDLAARNVLVRTFQ. -D------------------ASKNCIERDVAARNVLLTSG . EAKNIIERDUKSNNIFLBEGL.. E A K N I I B R D U K S N N I F L B L . . ....TVRIGDffiLATVKSRWSGSQQVEQP%. ...SVWUAPeVIRnpDDN.. .PFSF6SDVYSYGIVLYBLMA.GELPYAEINNRDIjIIPUVGIY ESQSILELDLRPANILISEPD... ...V C K I S D f f i C S ~ K L Q D L R G ~ A S P P E I G G . . . T Y T E ~ A P E I L K G E . . . . . . I A T P ~ D I Y S f f i I T L W Q U T T . R E V P Y S G E F Q W P Y A W A Y N L R e

arr

..LVKIIDffiLARRYN.PNEKLKVN..ffiT..

...NIRLTDffiFSCPLD.PGERLREV..CCT ... .........YARLVDffiFAKRIG.FGKRIWTF. . .PE

EKHRVLELDLKPENILCVNE.. E~NIVBRDLRPENILLDDMI... ESKGIIYRDLKF%NLILDE1IC:. BSLLILIYRDLKPENLLIWQ. 8 eDLKPEN** d

I

NA.

ffiWSRQ....

S

R

C-

WLCK PhK-Y CGPR CAPR

m C -m

z

Abl

foa tk8 Lps

1NS.R

..VKDLVSRFLWQFQKRYTAEEALAEPFFQQYWEEVREFSPRG RFKVICLTVLASVRIYYQYRRVKPVTREIVIRDPYALRPLRRLIDAYAFRIYGEWV/358

HYPDEETWAVSDE YQffiSWWDDYSDT. DUIEFPRRIA.KN.. KVRFPSEFS SD. a

.

.AWWSNLIVREQGMWSAAQCL~~L~.WEMRRCNRRLKSOILLRRlLUKR~K~PIAVSMNRPKR~?)

N L I R R L C R D N P S E R L G N L ~ V K D I Q ~ R W P E G P N W ELRKGTLTPPI G IPSVASPTDTSNFDSPPEDNDE PPPDDNSGWDI .DF ( 6 7 0 ) KNAI(WFATTOWIAIYQRRAPF1PRFKGPGDTSNFDDYEEEEIRVSINEKCGREPSEF 349) . ..MLKDLLRNLlQMLTKRffiNLRVNDI dl R e

P

P cp RnPCPPeCPES.

(

K

sd

I

... E--I-kEDYPTSTEPQYQPENL (533) .L% LHRLCWRKDPDeRFl’FEYIQSFLEDY F l ’ M g P S Y 8121 ...LYBAPIEPZWRLDPEERFl’WYLPSPLEDYFNGPPPB(6631 SMLQAPe LPPIPTRTCRPAAEQKDAPDTFELLETKGLGESDALDSEP/687 .....WEL~~QWNPSDRPSFAEIEPAPETWPPESSISOEVEgELGRRGTRGGAG V Y R L ~ P R C W E Y D P E R R P ~ P A V ~ D L I A I R ~ E1R1(8 2 ) .VPRLPIEQCWAYBPGQRPSFSAIYQE~SIRKRER(958) IRDLHTRCWAQDPBNRPTPFYIQEKLQE IRESPLCFSYFLGDRE SVAPLRIQTAFmPL 552) ......VTDLHRMCWQFNPNHRIFLE IVNL~DLBPSFPEVSFFBSEENIU~SEELEPIE:PeD~NVPLDRSSECQREEPGGRM;GSSLGPLRSYEEEIPYTE/13Z4 .VYUI~NKCWUIMDSRPRPLIAEFSKEVLRDPPRYLVIQGDERnaLPSPTDSKPYRTLI(DeEDllBD1VDADEYLVPAQGFFNSPSTSSRTPLLSSLSA/467

..-..----------R--------

RIIPCWGCPES.... RIIPCPPGCPAS.. EMERFSGCPER.. RLEPPEQCPED.. RLPCPELCPDA.... RLESPNNCPDD.. YLDQPDNCPER.. RLWPPICTID....

(

(

EGF.R

Em

C -€ma

mil

Laf Ems C-

mu

~

~-~ ~

.

ASPDLSRLY KNCP RAII ( R L V A ~ ~ V R R \ I R RPLFPI~ EE ILSSIE LLQESLPKI NRSAPE PSLRPAAETEDINACTLTTSPRLPVF ( 7 10 J PSLAGAVPTASLTG KAMNX IQSCWEAK LQRPSAELLQRDLMFffiTLG ( 3 7 4 I

198


phosphorylation on pp60v-src.There is a major site of tyrosine phosphorylation at residue 416 (39, 40); other minor sites of tyrosine phosphorylation are located in the N-terminal half of the molecule but have not been mapped precisely (41-44). All these sites can apparently be phosphorylated by pp60v-srcitself in a so-called “autophosphorylation” reaction. Whether this is a true intramolecularreaction or not is unresolved. pp60v-srcis also phosphorylated on serine with the major site being Ser- 17 (36,40).This serine can be phosphorylated by the CAMP-dependent protein kinase in vitro ( 4 3 , and there is evidence that elevation of CAMPlevels in some RSV-transformed cell lines leads to increased N-terminal phosphorylation of pp60v-src(46),presumably at this site. Recently it has been shown that there are other sites of serine phosphorylation in the N-terminal half of the protein which can be phosphorylated by the Ca2 -phospholipid-dependentdiacylglycerol-activatedprotein kinase, protein kinase C (47,48).Ser- 12 and to a lesser extent Ser-48 are phosphorylated in pp60v-srcisolated from cells treated with phorbol esters, compounds which activate protein kinase C (48). In RSV-transformed chick cells it has been estimated that about 60% of pp60v-srcmolecules are phosphorylated at Ser-17, while 30% are phosphorylated at Tyr-416 (28). Upon treatment with phorbol esters at least 50% of pp60v-srcmolecules are phosphorylated at Ser-12 +

(48).

The effects of these phosphorylations on the catalytic activity of pp60v-srchave not been determined with certainty. Phosphorylation at Ser-17 may increase activity a few-fold (46),but deletion of this residue by site-directedmutation does not abolish either protein kinase activity or transforming ability (36).The effects of phosphorylation at Ser-12 have not yet been determined, but, by analogy with phosphorylation at the neighboring Ser- 17, might be expected to increase catalytic activity. Phosphorylation at Tyr-416 has also been reported to increase the activiiy of pp60v-src(49),but once again a mutant pp60v-srcin which Tyr-416 has been replaced with a phenylalanine is able to function as a protein kinase and can transform (50, 51). It appears that none of these phosphorylations is essential for activity but may provide a means of positive regulation. In most respects pp60c-srcis very similar to pp60v-src,except at its C-terminus. From residues 1 to 514 chicken pp60c-srcdiffers from pp60v-srcby amino acid substitutions in a few scattered positions, the exact number depending on the strain of RSV. Starting at residue 515, however, the proteins diverge completely (52); has 19 additional amino acids encoded by contiguous sequences beyond those for residue 514, thus generating a protein with 533 amino acids. In contrast the 12 C-terminal amino acids of pp60v-srcare derived from sequences about 1 kilobase downstream of the true C-terminus of the c-src gene as a consequenceof a deletion that occurred during the genesis of RSV (52). In primary fibroblasts and most cell lines pp60c-srcaccounts for about 0.005% of total cell protein, which is a considerably lower level than pp60v-srcin RSVtransformed cells. There are cell types, however, where pp60c-srcis more abun-


199

dant. Certain neural cell types have about five times as much pp60c-"rc as fibroblasts (53,5 4 , while platelets have even higher levels (55). The majority of pp60c-srcin rat neurons migrates more slowly than pp6OC-"" from rat fibroblasts on gel electrophoresis (53) and appears to be altered in its N-terminal half. The existence of a unique form of pp60c-src in neurons raises the possibility that multiple forms of pp60c-srccould be produced from a single gene, for instance by alternate splicing, as occurs for the c-abl gene. Like pp6OV-"", pp60c-src is modified by myristylation of its N-terminal glycine (8), and most of pp6OC-"'" appears to be membrane-associated in a fashion very similar to pp6OV-"" (21), although whether the microscopic distributionof the two proteins is the same is not certain. pp60c-"rcis also phosphorylated on both serine and tyrosine. Tyr 416, however, is not a major phosphorylation site in pp6OC-"", even though the sequence around Tyr 416 is identical to that in pp60v-src(52).Another tyrosine in the C-terminal half is phosphorylated, which has been shown to be Tyr 527 in the unique C. pp6OC-"" autophosphorylatesat Tyr terminal tail of pp60c-src( 5 5 ~ )Interestingly, 416 in vitro (39).Phosphorylation of Ser 17 in pp6OC-"" is observed in vivo (56), and this residue is phosphorylated in vitro by the CAMP-dependent protein kinase (57). The pp60c-srcis also phosphorylated by protein kinase C at Ser 12 (48). The effects of these phosphorylationson the catalytic activity of pp6OC-"'" are currently under investigation. There is mounting evidence that phosphorylation of Tyr 527 in pp60c-srcacts as a negative regulator of its protein kinase activity (58).The c-yes and c-fgr proteins have a tyrosine in a homologous sequence near their C-termini, and these enzymes may also be negatively regulated by phosphorylation at this site. Considerable effort has gone into trying to understand how pp60v-srccauses transformation. Because of the similarity between pp6OC-"" and pp6OV-"'" it was originally proposed that transformation might simply be due to the 10-50-fold greater levels of pp60v-srcthat are present in RSV-infected cells. This idea has now been rigorously tested by the expression of a cloned c-src gene in susceptible fibroblasts at levels equal to or greater than those usually achieved for pp60v-"rc (59-62). Such cells are at best partially transformed. Therefore it appears that there is a qualitative difference between pp6OV-"" and pp6OC-"". In v i m recombination experiments between cloned v-src and c-src genes show that the alteration at the C-terminus of pp60v-srcis a major factor responsible for the transforming ability of this protein (59),although there are ancillary effects of some of the point mutations in the rest of the protein. For instance the substitution of Thr-338 by Ile in the pp60c-srccatalytic domain is apparently sufficient to activate it for transformation (63). The exact consequences of the altered C-terminus for the function of pp6OV-"", however, are not known. The presence of large amounts of pp60c-srcin nondividing cells such as platelets and neurons suggests that pp60c-srcmay not normally be

200


involved in proliferation. Instead pp60c-srccould have a role in a membrane process such as secretion ( 6 4 ,which is a major activity in both platelets and nerve terminals. The apparently newly acquired ability of pp6OV-"" to stimulate cell growth could be accounted for if pp60v-srchad different substrate specificities to pp60c-src.In vitro, however, pp60v-"rcand pp6@-"" display similar abilities to phosphorylate a wide variety of proteins and peptides, albeit few of which are physiological substrates (65,66).The major differenceis that the specific catalytic activity of pp60c-srcis lower than that of pp60v-srcin immunoprecipitates(65,66). This activity difference is even more apparent in vivo. Cells expressing a level pp60c-srcequal to that of pp60v-srcin a v-src-transformed cell have a level of phosphotyrosine in protein the same as that in a normal cell (65),in contrast to the 10-fold elevation observed in a v-src-transformed cell. Thus the activity of pp60c-srcis severely restricted in the cell. To what extent this is due to posttranslational modification(s), a regulatory protein(s), or limited access to suitable substrates is unclear, although as previously indicated the phosphorylation of Tyr-527 could play a key role. Certainly there is a good correlation between the occurrence of phosphate at Tyr-416 in the cell and the ability of a pp60"" molecule to transform. This leaves open the question of whether pp60v-src has different substrate specificities in the cell compared to pp60c-src,or whether transformation is a consequence of the cell being unable to regulate the activity of pp60v-"rc. B. PgOg'g-Yes

AND THE

c-yes PROTEIN

The v-yes oncogene occurs in two distinct avian sarcoma viruses, Y73 virus and Esh sarcoma virus. In contrast to RSV, the v-yes sequencesare expressed as part of a chimeric protein with viral structural gag gene sequences at its N-terminus. This gag-yes protein is predicted to contain 220 amino acids from the gag gene, 585 from v-yes sequences, and 7 C-terminal residues from an unused reading frame in the env gene (67).The expected 90-kDa protein is immunoprecipitated from Y73 virus-infected cells with anti-gag protein serum (68).Such immunoprecipitates have protein-tyrosine kinase activity detectable both by autophosphorylation of P90gag-yes and phosphorylation of added substrates. There are currently no yes-specific antibodies, although some of the monoclonal antibodies against ~ p 6 0 ' - "(14) ~ ~ and the anti-src 498-5 12 peptide serum (69) immunoprecipitate P90gw-w. Studies on biogenesis of P90gag-yes have shown that it is not modified by attachment of a lipid moiety (70). Instead the N-terminus is probably acetylated, like that of its progenitor P r 7 6 g U g . P90gug-Yesis distributed in the cell in very much the same manner as pp60v-src, being largely membrane-associated (69). The nature of its attachment to membranes is not understood, but newly synthesized molecules of P90gag-yesare associated with pp89 and pp50 in a complex similar to that observed with pp60v-src(71, 72).


20 1

The location of the catalytic domain in P90gag-yeshas not been defined by isolation of proteolytic fragments, but there is a striking sequence homology between residues 555 and 801 of P90gag-Yesand residues 270-516 of pp6OV-"", - ~P90gag-yes ~~ also where over 90% of the residues are identical (Fig. 1). p ~ 6 0 "and have sequence homology upstream of the catalytic domain: greater than 80% of the residues between 365 and 554 of P9Wag-yesare identical to residues 81-250 of pp6OV-"". There is no homology, however, between the N-terminal membranebinding domain of pp6OV-"" and P90gag-ye".Despite this close relationship there is no doubt that the c-src and c-yes genes and their products are distinct entities. P 9 0 g a g - y e s contains two sites of tyrosine phosphorylation and two sites of serine phosphorylation (73). One of the serine sites is in the gag region, and presumably corresponds to the site in p19gag. Tyr-700, homologous to Tyr-416 in pp60v-src, has been identified as one of the sites of tyrosine phosphorylation and is the major site of autophosphorylationin v i m (40). The effects of these phosphorylations on P90g'g-Yes protein kinase activity are unknown. The product of the c-yes gene has not yet been identified, although transcripts of the gene are abundant in kidney and a variety of embryonic tissues (74). Because there are 585 v-yes-encoded amino acids in P90gag-yes and because the v-yes sequence is apparently truncated at both its N- and C-termini, it seems likely that the c-yes protein will be larger than 65 kDa. No mutations have yet been made in the v-yes gene to define which sequences are critical for transformation, and to determine how the c-yes gene has been activated. It is possible that, as is the case for pp60v-src,the alteration at the C-terminus of P90gag-yes vis-A-vis the c-yes protein may be important.

C. P7Wag-fgr AND THE c-fgr PROTEIN The v-fgr gene is carried by the Gardner-Rasheed (GR) feline sarcoma virus. The transforming protein of GR-FeSV, P70gag-fgr, is in fact a tripartite protein, predicted to have 663 amino acids of which, starting at the N-terminus, 118 are from the gag gene, 151 are from the 5'-end of a mRNA-encoding y-actin (including the 128 N-terminal amino acids), 390 are from v-fgr sequences, and 5 are from the parental feline leukemia virus env gene in an unused reading frame (75). The 390 v-fgr-encoded amino acids are extremely homologous to the Cterminal region of the v-yes gene, but the v-fgr and v-yes genes are apparently derived from two distinct cellular genes (76). A protein of 70 kDa, which has a tyrosine-specific autophosphorylating activity, can be immunoprecipitated from GR-FeSV transformed cells using anti-feline p15gag serum (77). There are no fgr-specific antibodies available. Little work has been done on the biosynthesis of P70gag-fRr,but it is modified by an N-terminal myristyl group, as is the feline gag gene precursor protein with

202


which it shares N-terminal sequences (78). P7OgUg-rSrhas been detected in the cytoplasm of transformed cells by immunofluorescence staining but there is no obvious coincidence of staining with membranes, despite the presence of the myristyl group (79). Subcellular fractionation studies suggest that P70gag-fgr is partially particulate and partially soluble. By analogy with pp60v-srcthe catalytic domain of P70g'g-fgr would be comprised of residues 407-516 (see Fig. l), but this has not been tested experimentally. Like P9OgUg-yes,P70gag-fgrshares sequences with pp60v-srcupstream of the catalytic domain. P7OgUg-fgrcontains three sites of tyrosine phosphorylation and two sites of serine phosphorylation (78). Tyr-553, located in a position homologous to Tyr-416 in pp60v-src,appears to be one of the sites of tyrosine phosphorylation and is the major site autophosphorylated in vim. There are no indications whether any of these phosphorylations affect the activity of P70gag-fgr. The c-fgr protein has not been identified nor have any mutations of the v-fgr gene been created. The intriguing question of whether the actin sequences in P7o8"g-fer play a role in its transforming activity remains unanswered. It is curious that the 3' recombination sites for the v-yes and v-fgr genes are identical with respect to the acquired cellular sequences, although the recombination sites in the parental viruses are different. Thus both proteins are apparently truncated with respect to their cellular counterparts, but have different C-termini. This again implies that precise loss of C-terminal sequences may be important for activation.

D. P140gag-fps AND p98c-fps; pg5sag-fes AND p92c-fes The v-fps oncogene is found in a series of avian sarcoma viruses, including Fujinami sarcoma virus (FSV), PRCII, and PRCIV. These viruses contain overlapping v-fps sequences all of which have a common C-terminal region of 430 amino acids but which vary to some extent to the N-terminal side of this region due to deletions in the v-fps sequences. The different viruses induce distinguishable transformed phenotypes, probably due to these variations. All the viral transforming proteins contain N-terminal viral gag gene sequences joined to v-fps sequences. As an example, FSV P140Sag-fpsis a protein with 308 gag-derived amino acids and 874 v-fps-encoded amino acids (80). The v-fes oncogene exists in two feline sarcoma viruses, Snyder-Theilin (ST) FeSV and Gardner-Arnstein (GA) FeSV. The two v-fes genes also have a common C-terminal region of 436 amino acids, and differ in the nature of the v-fes sequences nearer their N-termini. These sequences are expressed as gag-fes chimeras, and, for example, GA-FeSV P1 10n'glfes has 425 gag-derived amino acids and 609 fes-encoded amino acids (81). There proved to be striking sequence homologies between the v-fps and v-fes genes and their products suggest-


203

ing that they were derived from genes that are equivalent in chicken and cats; this has been formally demonstrated (82).As a result one would expect the v-fps and v-fes proteins and their cellular counterparts to be similar in their properties. To a large extent this has turned out to be the case, and we therefore discuss thefps and fes gene products together. Proteins of the expected sizes can be immunoprecipitated from v-fps-transformed cells either with anti-p19gag serum or anti-fps tumor serum (83, 84). There are also monoclonal antibodies that recognizefps determinants mapping to a region N-terminal to the protein kinase domain (85). Immunoprecipitates of P140gag-fps have tyrosine-specific autophosphorylating activity and can phosphorylate exogenous substrates (86, 87). The gag-fes proteins can be immunoprecipitated from v-fes-transformed cells either with anti-feline p15gag serum (88, 89) or anti-fes tumor serum (88). There is also a series of anti-fes peptide sera (90). In immunoprecipitates, gagYes proteins will autophosphorylate and phosphorylate added substrates on tyrosine (88, 89). Metabolic labeling studies have shown that newly synthesized P140k'ug-fps molecules are found in association with pp89 and pp50 (71, 72). Since the Nterminus of P140gag-fpsis derived from the avian gag gene precursor, Pr76gag, in all likelihood it is acetylated although this has not been formally shown. Halflives of 3-5 h have been measured for several of the different gag-fpsproteins (70).Unlike pp60v-srcand P90gag-yes, P140gag-fPsis broadly distributed in the cytoplasm showing a diffuse staining pattern (92, 92).The absence of membrane association is corroborated by cell fractionation studies, where P140gag-fPs behaves as a soluble protein at physiological ionic strength (91). P140gag-fps, however, is found in the particulate membrane fraction at low ionic strength suggesting it does have some affinity for membranes (92). In contrast to the gag-fps proteins, the N-termini of the gag-fes proteins are myristylated ( 9 3 , like the feline gag gene precursor from which they stem. By imrnunofluorescence staining P85gag-feshas been found in the cytoplasm as well as in association with membrane structures (94).The myristyl group may predispose gag-fes proteins to this more pronounced membrane affiliation. The catalytic domain of P140gag-fpshas been defined by analysis of proteolytic fragments and lies at the C-terminal end of the protein (95). Comparison of the sequence of P140gag-fps with that of pp60v-srcshows that residues 924-1 174 of the v-fps sequence have striking homology with the catalytic domain of pp60v-src (residues 270-516) (Fig. 1). This region is essentially identical in all the strains of virus carrying the v-fps gene. Residues 700 to 950 form the equivalent catalytic domain in PI 10gag-fes(Fig. 1). There is also evidence for a second functional domain in P 140gag-fps as discussed below. P140gag-fps contains single minor sites of both serine and tyrosine phosphorylation in the gag-derived region, as well as one serine and two tyrosine sites in the v-fps region (95). The only identified site is Tyr 1073, which is the

204

TONY HUNTER AND JONATHAN A . COOPER

tyrosine homologous to Tyr-416 in pp60v-src. Tyr-1073 is one of the major acceptors for the in vitro autophosphorylationreaction. Mutation of Tyr- 1073, to Phe, Ser, Thr, and Gly has been accomplished (96, 97). In each case the mutant shows a diminished ability to phosphorylate added substrates and is noticeably weaker in its transforming activity. One presumes that these defects are due to the inability to phosphorylate this residue, although there is an outside chance that a tyrosine per se is required at this site for full activity. At the very least Tyr-1073 appears to be more important than Tyr-416 in p ~ 6 0 ' - " ~There ~. are preliminary indications that phosphorylation at this site increases the phosphotransferase activity of P140g'g-fis. Both the c-fps and c-fes proteins have been identified using a variety of antiv-fps and anti-v-fes antibodies. The c-fps protein is a 98-kDa polypeptide which is found predominantly in hematopoietic cells (98). The p98'-fPs has proteintyrosine kinase activity in v i m , both for autophosphorylation and towards exogenous substrates; ~ 9 8 C - happarently ~ lacks phosphotyrosine when isolated from cells, but contains phosphoserine. In contrast to P140g'g-fps, p98"-fiS is preodminantly soluble upon cell fractionation even at low ionic strength (99). The c-fes protein, somewhat smaller at 92 kDa, is also found in hematopoietic cells, particularly of the myeloid lineage (100,101). Like p98'gPs, p92'-feS autophosphorylates and phosphorylates added substrates. Interestingly a 94 kDa protein with tyrosine-specific autophosphorylating activity, is also found on immunoprecipitation of mammalian cells with rat anti-v-fps tumor serum (100) or anti-fps peptide serum (101). The distribution of this protein in different cell types is distinct from that of p92"-fes, and their relationship is unclear. Although much of the sequence of the c-fps protein has been deduced from the nucleotide sequence of chicken c-fps genomic clones (102), the entire sequence of neither p98"-fPs nor p92"-feSprotein is known. From the fact that the v-fps sequences in FSV encode approximately 95 kDa, however, it is likely that they represent most of ~98~-fi". The requirements for the transformation byfps sequences have been studied in detail. The gag sequences have been shown to be dispensible for transformation (103), although such mutant viruses are more weakly tumorigenic, as are viruses like PRCII which lack N-terminal v-fps sequences (104, 105). A virus in which cellular sequences homologous to the v-fps sequences in FSV have been substituted for the v-fps sequences is capable of transformation (106). The gag-derived sequences, however, are necessary for transformation by c-fps sequences (106). This suggests that there are subtle differences between v-fps and c-fps necessary for the transforming ability of v-fps, which remain to be determined. For instance there are at least 26 single amino acid differences between the c-fps protein and the v-fps protein in the shared region (102). The C-termini of the v-fps and c-fps proteins are the same, and so, in contrast to ppf50v-src,there is no indication that changes at the C-terminus are required. Linker insertion mutations


205

have been created in many positions of the FSV g a g - . s gene (107). Mutants with insertions into the catalytic domain are invariably nontransforming, as are those with insertions into the 100 residues upstream of this domain. Such mutations also define a region between residues 400 and 600 of P140gag-fpSwhich is necessary for transformation. Mutants with insertions in between these two regions are fully active, suggesting that P140gag-fps may have a two-domain structure (108). A start has been made with identifying thefes sequences required for transformation; substitution of 80% of the v-fes sequences in GA-FeSV with the equivalent human c-fes sequences reduces but does not abolish transforming activity (109).

E. P 120g'g-abl

AND

p 150c-ab'

The v-abl gene was initially identified in Abelson murine leukemia virus (AbMuLV), but has subsequently been found in HZ-2 FeSV. Ab-MuLV induces B cell disease in mice, but can transform many other types of cell, including fibroblasts, under the right circumstances. P160gag-ab' contains 236 gag genederived amino acids at its N-terminus, followed by 1008 v-abl-encoded residues (110, 111). P160eag-ab' can be detected in Ab-MuLV-transformed cells by immunoprecipitation with either anti-murine P15g'g serum (112, 113) or an unusual type of antitumor sera from a C57-leaden mouse (114). Immunoprecipitates of P160e'g-ab' made with anti-gag sera display autophosphorylatingactivity and can also phosphorylate added proteins (115, 116). Recently several site-specific antibodies have been generated, both against bacterially expressed fragments of the v-abl sequence and against synthetic v-abl peptides (117, 118). The antibodies against sequences derived from the protein kinase domain inhibit the P160gag-ub'protein-tyrosine kinase activity, while those against sequences outside the protein kinase domain have no effect (117). Variants of Ab-MuLV encoding smaller gag-abl proteins have arisen at a high frequency. A commonly used strain encodes P120gag-a6' which has an internal deletion in the C-terminal half of the protein (110, 111). A catalytic domain can be defined in P160s'g-ub' by comparison of its sequence with that of pp60v-Src.Residues 367-614 of P160gag-ab[have strong homology with the catalytic domain of pp60v-"rc(Fig. 1). The positioning of this domain near the N-terminus of the v-abl sequences distinguishes P160gag-ab' from the viral protein-tyrosine kinases discussed so far. An identical location for the catalytic domain has been determined experimentally by expression of fragments of the v-abl gene in E. coli and assaying the resultant abl proteins for protein-tyrosine kinase activity (119). As will be seen in the final paragraph of this section, much of the sequence C-terminal to the catalytic domain is dispensible for transformation. P160n'g-ab' is myristylated like Pr65gag (93), the gag gene precursor of

206


MuLV. Mutants in which the N-terminal region of P160g'g-ab' has been exchanged for one which cannot be myristylated are nontransforming suggesting that this modification is critical for transformation (120). Immunofluorescence staining localizes P 160gag-ab'in the cytoplasm of transformed fibroblasts, where it is associated with the plasma membrane and concentrated in adhesion plaques (121). An early study with antitumor serum apparently detected P160gag-ab'on the surface of transformed cells (114), but later-evidence suggests that the fluorescence may not have have been due to P160g"g-ab'. Cell fractionation studies show that P160gag-ab'is associated with particulate fractions and the cytoskeleton, but there is also some soluble protein (122). P16CEag-ab'is multiply phosphorylated and contains four sites of serine phosphorylation, two of threonine phosphorylation, and two of tyrosine phosphorylation (123). Two of the serine sites are located in the gag-derived region. Tyr-514, one of the sites of tyrosine phosphorylation (40, 124), is homologous to Tyr-416 in pp60"-src. The other site of tyrosine phosphorylation may correspond to Tyr-385 (40, I 1 7). In vitro P160eag-ab' undergoes autophosphorylation at multiple sites. One of these is Tyr-514, but the other major sites are not detected in P160gag-ab' labeled metabolically (123, 124). The functional effects of these phosphorylations have not been ascertained. The product of the murine c-abl gene has been identified as a 150-kDa protein (125).Recent evidence obtained by sequencing c-abl cDNA clones suggests that there are at least four distinct c-abl proteins which share their C-terminal 1010 residues but which differ at their N-termini, containing 20-40 unique amino acids as a result of the use of alternate 5' exons in their mRNAs (126).p150c-ab' is found predominantly in cells of hematopoietic origin, but occurs at low levels in many cell types. The product of the human c-abl gene ( ~ 1 4 5 ~ - " ~which ' ) , is slightly smaller, also occurs in hematopoietic cells (127). Apparently neither protein contains phosphotyrosine, although both are phosphorylated on serine (127, 128). Originally p150c-ab' was reported to lack protein-tyrosine kinase activity (128), but by altering assay conditions it has been possible to show that p150c-ab' both autophosphorylates and phosphorylates added substrates (129). The major site of autophosphorylation corresponds to Tyr-5 14 in P160gag-ab' (120). Little is known about the function of p150c-ab'but the c-abl gene is found to have undergone a specific type of rearrangement in the leukemic cells of the majority of patients suffering from chronic myelogenous leukemia (130, 131), which results from a reciprocal translocation of part of chromosome 9 onto chromosome 22. In the translocation the c-abl gene on chromosome 9 is truncated at its 5' end and comes under the control of a transcription unit on chromosome 22 termed bcr. This results in a novel hybrid bcr-abl mRNA which generates a 210-kDa protein containing bcr sequences at its N-terminus followed by c-abl sequences (118, 127). The p210bcr-ab'is active as a protein-tyrosine kinase


207

in v i m and is phosphorylated on tyrosine in the cell (127). Presumably the enzymic activity of p2 10bcr-ab/is important in the leukemic phenotype, but substrates specific for this protein-tyrosine kinase are still being sought. The sequence requirements for transformation by v-abl have been rigorously analyzed. All except the N-terminal 14 amino acids of the gag-derived sequences have been deleted, leaving sufficient information to dictate myristylation (120, 132). This gag-deletion mutant has undiminished fibroblast transforming activity, but decreased ability to transform lymphoid cells. This decrease, however, is probably due to instability of the protein in lymphoid cells rather than an inherent deficiency (133). Provided there is an N-terminal sequence with a myristylation signal, the minimum region required for transformation of fibroblasts has been defined as residues 250-630 of P160g'g-ub' (134). This is somewhat longer than the minimal protein kinase domain. The additional residues are on the N-terminal side of this domain in a region of homology with pp60v-srcand P140g'g-fps. Although the residues from 630 on can be deleted without loss of fibroblast transforming activity, such mutants have attenuated ability to transform lymphoid cells (132). The structures of the oncogenically activated abl proteins, P160g'g-ab' and P210bcr-ub',are in many ways similar in that both have lost N-terminal sequences present in p150C-ub'.When isolated from the cell the specific activity of the p150c-ub'protein kinase appears to be lower than that of either P160g'g-ub/ or p210bcr-ub'.This suggests that the N-terminal sequences of p150c-ub'may act as a negative regulatory domain. This idea is supported by the fact that there are different species of p150c-ub'with different N-terminal regions (126, 1 3 9 , each of which could have a specific regulatory function. Elimination of this regulatory domain, like that of the C-terminal domain of pp60c-src, may be critical for activation of the c-abl gene. This hypothesis remains to be tested directly. F. P68gag-rosAND THE c-ros PROTEIN The v-ros gene is carried by the UR2 avian sarcoma virus. The predicted structure of the v-ros gene product has 150 gag gene-derived amino acids at its N-terminus followed by 402 ros-encoded residues (136). A protein of 68 kDa can be immunoprecipitated by anti-p 19g'g serum from UR2 virus-transformed chick cells (137). In such immunoprecipitates P68gag-rosautophosphorylates on tyrosine (137). One ros-specific antitumor serum has been generated (138). P68gag-rosis not myristylated (138), but residues 158-175 are uncharged and could form a membrane attachment site. Immunofluorescence staining shows is present in the cytoplasm of transformed cells, but no prothat P68gng-rcJS nounced concentration of the protein in adhesion plaques or cell junctions was noted (139). Cell fractionation studies, however, suggest that P68g'g-roS is plasma membrane associated (136).

208


Comparison of the sequence of P68gag-roswith that of pp60v-srcdefines residues 251-518 as the catalytic domain of P68gUg-'Os (Fig. 1). The sequences on either side of this catalytic domain are relatively short and therefore probably not long enough to form separate functional domains, although the putative membrane attachment site could be critical for transformation. pfjggag-r~s is phosphorylated in transformed cells predominantly on serine (137). Efforts to detect phosphotyrosine in P68gag-'Os have been unsuccessful. Autophosphorylation of P68gag-'OS occurs at a single major site. The identity of this tyrosine is unknown, but the properties of the phosphotyrosine-containing peptide are not consistent with its being Tyr-418, which is the tyrosine homologous to Tyr-416 in pp60v-src (140). In contrast to the viral protein-tyrosine kinases discussed so far P68gUg-'OSdoes not induce dramatic changes in the level of phosphotyrosine in proteins in transformed cells (141, 142). P68g'g-r0s may therefore have a more restricted substrate specificity than the other viral proteintyrosine kinases. The product of the c-ros gene has not been identified, although a 3.1 kb c-ros transcript has been detected exclusively in kidney (143). The sequences of the chicken c-ros gene and the v-ros gene diverge at residue 540 of P68g'g-ros just downstream of the protein kinase domain. In this respect P68g'g-r0S is similar to pp60v-src,P90g'g-Yes, P70gag-fsr, and gP180g'g-fms (see Section 11, H). Among the protein-tyrosine kinases the sequence of the catalytic domain of P68g'g-r0s is most closely related to that of the insulin receptor (see Fig. 1). This coupled with a second structural analogy to the EGF and insulin receptors, namely the existence of a putative transmembrane domain upstream of the protein kinase domain, strongly suggests that the c-rm protein will be a cell surface receptor with a ligand-regulated protein-tyrosine kinase activity. The nature of this receptor is of obvious interest.

G. g ~ 6 8 / 7 2 ' - ' ~ ~AND -~

THE

EGF RECEPTOR

The v-erb-B oncogene is found in two distinct avian erythroblastosis viruses, AEV-ES4 and AEV-H. The former carries an additional and unrelated cellderived oncogene, v-erb-A, which potentiates the effect of the v-erb-B gene. Although in inefected animals these viruses cause erythroblastosis, in culture both viruses can transform fibroblasts. The v-erb-B gene is of particular interest because it was derived from the chicken EGF receptor gene, of which it represents a part (see below). The complete sequence of the AEV-H v-erb-B gene has been determined ( 1 3 3 , but the true N-terminus of the v-erb-B protein is unknown. The v-erb-B gene is expressed from a subgenomic RNA and therefore the N-terminus of the v-erb-B protein could either contain 6 amino acids of the gag gene product or be initiated within v-erb-B sequences. Assuming the former


209

to be correct, then the v-erb-B protein has 6 gag-gene-encoded residues plus 609 from v-erb-B, predicting a protein of 61 kDa (144). In fact cells transformed by AEV contain multiple forms of v-erb-B protein (145, 146), which appear to represent differently glycosylated forms of a single polypeptide chain of about 61 m a . The major v-erb-B gene product is g~66/68'-"'~-~. A small fraction of g ~ 6 6 / 6 8 ' - " ~is~further -~ modified to give proteins of 75-80 kDa ) (147-149). The various v-erb-B proteins can be immunoprecipitated with antitumor serum raised in rats ( 1 4 3 , or by antisera raised against a bacterially expressed protein corresponding to a fragment of the v-erb-B protein (146).A number of anti-erb-B peptide sera have also been reported (150, 151). All the different forms of v-erb-B proteins are found in cellular membranes with which they become associated as a result of synthesis on membrane-bound ribosomes. It is unclear if a cleavable signal peptide is necessary for this interaction, since, as previously explained, the true N-terminus of the v-erb B proteins is unknown. Neither g ~ 6 6 / 6 8 ' - ' ~ ~ nor - ~the EGF receptor are modified by attachment of lipid. Assuming there is a gag-erb-B chimeric protein, and numbering from its N-terminus, the following structure can be deduced for g ~ 7 4 ' - " " ~ -The ~. first 74 amino acids (perhaps lacking those residues upstream of a signal peptide cleavage site) would lie outside the cell. This region contains three sites for Nlinked glycosylation, and there is evidence for the presence of three oligosaccharide chains on g~66/68'-"'~-~. Residues 75-99 are uncharged and would constitute a transmembrane domain anchoring the protein in the membrane. Residues 100-615, which would then lie in the cytoplasm, include a protein kinase domain starting about 50 residues from the end of the transmembrane domain. Cell fractionation studies show that g ~ 6 6 / 6 8 ' - " ~is- ~associated with intracytoplasmic membranes which appear to be perinuclear and may represent part of the Golgi apparatus (147, 148). In contrast g ~ 7 4 ' - " ~ -is~largely exposed on the cell surface, and can be detected by immunofluorescence staining (147). From the slow rate of maturation of g ~ 6 6 / 6 8 " - " ~it- 'seems ~ as if the v-erb-B gene product is delayed in its transit through the Golgi apparatus. However, neither transport of the v-erb protein to the cell surface nor transformation is affected by the presence of the N-linked glycosylation inhibitor, tunicamycin. Residues 147-399 of the predicted v-erb-B gene product show strong homology with the catalytic domain of p ~ 6 0 " (Fig. - ~ ~l), ~ yet initial attempts to demonstrate protein kinase activity associated with v-erb-B proteins were unsuccessful (145, 146). This was doubly surprising in light of the subsequent finding that the v-erb-B protein represented a part of the EGF receptor gene (152), whose product has bona fide protein-tyrosine kinase activity that is stimulated upon binding EGF (153). Recently, however, protein-tyrosine kinase activity has been detected in immunoprecipitates of v-erb-B proteins, by autophosphorylation and phosphorylation of exogenous substrates (149, 150), as well as in membranes from AEV-infected chick cells (154). The reason for this discrepancy is unclear;

210


some of the antibodies used in the earlier experiments may have inhibited the protein kinase activity. In addition the levels of in vitro phosphorylation are low by comparison with the other virally coded protein-tyrosine kinases and by comparison with the EGF receptor. The v-erb-B proteins are only weakly phosphorylated in either AEV-transformed fibroblasts or erythroblasts (145, 149), containing predominantly phosphoserine and some phosphothreonineat sites which are unidentified. and g ~ 7 4 ' - " ~may - ~ be substrates for protein kinase C, since all three forms of the v-erb-B protein show increased phosphorylation in response to 12-0-tetradwanoylphorbol- 13-acetate(TPA) treatment of AEV-infected chick fibroblasts (149).One site of protein kinase C phosphorylation may be Thr-109, since the equivalent Thr in the EGF receptor (Thr-654) is phosphorylated by this enzyme (155, 156). Phosphorylation of Thr-654 diminishes the degree of stimulation of the EGF receptor protein-tyrosinekinase by EGF and decreases the affinity of the receptor for EGF (157, 158).The effect of the TPA-induced threonine phosphorylation on the v-erb-B protein-tyrosine kinase has not been determined, although TPA treatment of one AEV-transformed fibroblast clone causes growth arrest (149). There is every reason to believe that the c-erb-B gene is the EGF receptor gene. The properties of the EGF receptor are reviewed in Chapter 7 by Morris White and C. Ronald Kahn. We discuss here only the features of the EGF receptor that are salient for an understanding of how the c-erb-B gene might be activated to transform. The complete sequence of the human EGF receptor has been deduced from the nucleotide sequence of overlapping cDNA clones (159). The EGF receptor has 1186 amino acids consisting of a 619-residue EGF-binding extracellular domain and a 542-residue intracellular domain separated by a single 26-amino acid transmembrane segment. The AEV-H v-erb-B protein corresponds to residues 551-1154 of the receptor, and thus is missing most of the extracellular domain (144). The v-erb-B protein is also truncated at the Cterminal end and lacks the last 32 residues of the EGF receptor. This region of the EGF receptor contains a major autophosphorylationsite (Tyr-1173) which is thought to be important in the regulation of its protein-tyrosine kinase activity (160). The absence of this acceptor site may in part explain why the in vitro protein kinase activity of the v-erb-B protein was hard to detect. A priori one might imagine that the loss of both the regulatory ligand-binding domain and the autophosphorylation site would lead to the constitutive activation of the v-erb-B protein as a protein kinase. In fact its specific activity seems to be lower than that of the unstimulated EGF receptor. This is borne out by the observation that AEV-transformed fibroblasts display only a very modest increase in the level of phosphotyrosine in protein (154, 161). Nevertheless the idea that the v-erb-B proteins mimic an occupied EGF receptor and deliver an unregulated growth signal is still very attractive. In keeping with this notion at least some of the phosphotyrosine-containing proteins unique to AEV-trans-


21 1

formed fibroblasts are also observed following EGF treatment of fibroblasts (see Section V,B). An indication that the expression of gp74V-erb-B on the cell surface is critical for transformation comes from the properties of temperature-sensitive mutants of AEV (147). Mutant AEV-infected erythroblasts revert to a normal phenotype upon shifting to the restrictive temperature and this is accompanied by a loss of from the surface. A start has been made in defining the regions of the v-erb-B gene that are necessary for transformation of both fibroblasts and erythroblasts, as well as delineating which changes in the c-erb-B gene are required for activation. Deletions and insertions in the protein kinase domain of the v-erb-B gene abolish transforming activity (162-164). A two-amino acid insertion in the connecting region between the protein kinase domain and the transmembrane domain is without consequence (164). Likewise large deletions downstream of the protein kinase domain (144, 163) and a smaller deletion or insertions into the extracellular domain have no effect on transforming ability (164). Starting with the intact EGF receptor, truncations at both the N-terminus and C-terminus are required for the protein to have transforming activity for fibroblasts, although exactly how many amino acids have to be removed at either end has not been determined (165). In contrast the normal C-terminus may serve for erythroblast transformation, since the c-erb-B gene in transformed erythroblasts, which has been activated by the insertion of an avian leukosis virus genome, has an intact C-terminus (166). The temperature-sensitive mutants of AEV may also be instructive in this regard. For instance there are two mutations in the ts34 AEVES4 mutant both of which lie within the protein kinase domain (167). The change of a histidine (equivalent to His-270 in the AEV-H v-erb protein) to an aspartic acid is likely to be the crucial one. This mutation has a clear-cut effect on transport of the protein to the surface at the restrictive temperature, yet, despite its location in the protein kinase domain, the protein-tyrosine kinase activity of the ts34 v-erb protein does not appear to be temperature sensitive.

H.

gP180g‘g-fms

AND

gp17OC-f””

The v-fms oncogene was originally characterized as the oncogene of the McDonough strain of feline sarcoma virus (SM-FeSV). It has subsequently been detected in a second feline sarcoma virus (HZ-5 FeSV). The nucleotide sequence of SM-FeSV predicts a chimeric product of 1511 amino acids, of which 536 are derived from the gag gene and 975 from the v-fms sequence (168). Several v-fms glycoproteins are generated from a single primary translation product (169-1 71). The first detectable product is a 180-kDa glycoprotein (gP180g‘g-f”“), which is processed to yield P6Wg and gpl2OV-fms(172, 173). The oligosaccharide side chains of the latter are then modified to yield a mature protein (gpl4OV-r””).All the proteins containing v-fms-encoded sequences can be immunoprecipitated

212


with rat antitumor serum (171, 173). Rat monoclonal antibodies to v-fms have been reported and these like the antitumor serum recognize extracellular v-fms domains (172, 174). Anti-gag protein sera will immunoprecipitate gP180gag-fms (169-173). Antisera against synthetic v-fms peptides and bacterially expressed proteins corresponding to v-fms have become available (1 75). gpl2OV-fmsis associated with intracellular membranes, while gp14OV-fmsis principally detectable on the cell surface both by immunofluorescence staining and by iodination (173, 176). The majority of gpl4OV-fmsappears to be located in coated pits (176). Since all the v-fms proteins are membrane-associated and indeed span the membrane, the N-terminus of the protein must provide a signal sequence. It is presumed that the synthesis of the gag-fms precursor is initiated at first in phase AUG codon which is normally used for the synthesis of the FeLV glycosylated gag gene product. The maturation of gpl2OV-fmsto gp140v+msis rather slow and incomplete (173), and, as is the case for the v-erb-B proteins, it appears that transit through the Golgi apparatus is impeded. Experiments with inhibitors of glucosidase I, which is involved in oligosaccharide maturation in the Golgi apparatus, show that maturation is required both for transport to the cell surface and for transformation (177). The gag-fms protein is modified by addition of myristate. This suggests that the signal peptide for insertion of the gag-fmsprotein into the membrane is internal and may be that normally used by p17OC-fms. The structure of the v-fms proteins is reminiscent of that of the v-erb-B proteins. There are 1080 residues, containing 14 sites for N-linked glycosylation, upstream of a 26-residue transmembrane domain. Fifty residues beyond the transmembrane domain recognizable homology with the catalytic domain of pp60v-srcis evident. Residues 1152-1478 are homologous to residues 270-516 of pp60v-src(Fig. 1). Note that this v-fms catalytic domain is 80 residues longer than that of pp60v-src.This is largely due to a 68-amino acid insertion between residues 358 and 359 of pp60v-src.An initial report that gp 12OV-fmsunderwent autophosphorylation on tyrosine (178), was later confirmed (1 79). Like the v-erb-B proteins, the specific activity of the v-fms protein-tyrosine kinase appears to be low, and cells transformed by the v-fms oncogene show no increase in the level of phosphotyrosine in cellular protein (178). When isolated from SM-FeSV transformed cells gp12OV-fmswas found to be phosphorylated on serine and threonine, but no phosphotyrosine was observed (178). The sites of phosphorylation have not been identified, but presumably lie in the cytoplasmic domain. The sites of autophosphorylation have not been mapped either. There is no indication whether phosphorylation of gp120gag-fms affects its activity. Given the structure and properties of the v-fms proteins it seemed likely that the product of the c-fms gene would prove to be a growth factor receptor. This


213

prediction has been substantiated by the finding that gp17OC-f””is the receptor for a hematopoietic growth factor-colony stimulating factor 1 (CSF-1)-which stimulates the growth of cells in the myeloid-monocyte lineage (175). This identification is consistent with the nearly exclusive expression of c-fms mRNA in hematopoietic tissues, and the demonstration that a 170-kDa protein can be immunoprecipitated from cat spleen with anti-fms monoclonal antibodies, which becomes phosphorylated on tyrosine in v i m (180). Furthermore tyrosine phosphorylation is stimulated by the addition of CSF-1 to membranes prepared from cell lines expressing CSF-1 receptors (181). The requirements for v-fms transformation have been investigated by sitedirected mutagenesis. A deletion just upstream of the transmembrane domain abolishes transforming activity even though the mutant protein retains proteintyrosine kinase activity (179). In the mutant-infected cells gpl 2OV-fms accumulates in the Golgi apparatus, and no gp14OV-f””is detectable. This implies that the expression of gpl4OV-fmson the cell surface is required for transformation, in keeping with the results obtained with glucosidase I inhibitors (177). A deletion which removes the transmembrane domain is also nontransforming (174). This mutant v-fms protein does not reach the surface either, and leaves no sequences exposed in the cytoplasm, apparently because the absence of the hydrophobic transmembrane domain allows transfer of the whole protein into the lumen of the endoplasmic reticulum. Deletion mutations in the protein kinase domain abrogate transforming activity (182). The requirements for activation of the c-fms gene have not yet been determined, but the v-fms and c-fms genes have different C-termini with the sequences diverging at residue 1498 of v-fms (183). From this point the c-fms protein has a further 39 amino acids while the v-fms protein has 13. By analogy with pp6OV-”“and pp6OC-”“this could prove to be an important difference. The v-fms protein apparently contains the normal c-fms N-terminal sequence (183), so it is unclear whether there are changes at the N-terminus of the gene, such as the addition of gag sequences, which are necessary for transformation. There are other scattered differences between the predicted v-fms and c-fins amino acid sequences, which might also be important. With regard to the mechanism of transformation, while all the available evidence indicates that the protein-tyrosine kinase activity of the v-fms proteins is rather restricted, the results with site-directed mutations suggest that this activity is critical; gpl4OV-fmsmust be expressed on the surface, and its localization in coated pits (176) suggests that it is mimicking a growth factor receptor. It seems likely that gpl4OV-fmsis constitutively activated as a protein-tyrosine kinase, since, although it can still bind CSF-1, CSF-1 does not stimulate autophosphorylation (182). It is interesting to note that the v-fms oncogene causes primarily sarcomas, whereas its cellular cognate regulates the growth of hematopoietic cells. Conversely, the v-erb-B oncogene transforms erythroid cells,

214


while the EGF receptor acts in nonhematopoietic cells. The identification of the substrates for viral and cellular forms of these two protein-tyrosine kinases may yield a clue to this apparent paradox.

111.

Other Transformation-Related Tyrosine Phosphorylation Systems

A. THEv-sis ONCOGENE, PDGF, AND THE PDGF RECEPTOR The v-sis oncogene was initially found in simian sarcoma virus (SSV) and later in the Parodi-Irgens feline sarcoma virus (PI-FeSV). The v-sis gene of SSV has been intensively investigated, and indeed yielded the first direct connection between viral transformation and growth control, when it was discovered that the predicted sequence of the v-sis gene product was highly homologous to that of the B chain of human platelet-derived growth factor (PDGF) (184-186). The v-sis gene of SSV is expressed from a subgenomic mRNA in the form of an envsis chimera containing 38 amino acids from the N-terminus of the viral env gene and 220 residues encoded by v-sis sequences. Homology with the B chain of PDGF starts at residue 99 and extends to residue 207. The retention of the Nterminus of the env gene supplies the env-sis protein with a signal sequence. The earliest detectable product of the v-sis gene is a 28-kDa protein which probably corresponds to the primary translation product minus the signal sequence with the addition of a single N-linked oligosaccharidechain at position 80 (186-188). The g~28'-"~" is rapidly dimerized and then is trimmed by proteolytic processing at both its N- and C-termini. At the N-terminus the cleavage site is probably the Lys-Arg sequence (residues 97 and 98), which would generate an N-terminus identical to that of the B chain of PDGF. The C-terminal cleavage site could be at residue 207, the residue at which most of the human PDGF B chain molecules end (189). There is evidence that the product of the v-sis gene is secreted and variable amounts of a 24-kDa dimer are found in the culture medium of SSV-transformed cells (190, 191). Both point mutations and small deletions within the signal sequence of the v-sis gene abolish transforming activity indicating that the v-sis products must enter the secretion pathway to be active (192). Since only cells bearing PDGF receptors can be transformed by SSV, there is every reason to believe that the v-sis proteins must function through interaction with the PDGF receptor, and thereby mimic PDGF. The PDGF receptor is a 180-kDa surface glycoprotein with an associated protein-tyrosine kinase activity which is stimulated severalfold upon binding PDGF (193-195). It seems likely that at least some aspects of the response of cells to PDGF are mediated through increased tyrosine phosphorylation of cellular proteins by activated PDGF receptor. This in


215

turn suggests that transformation by the v-sis oncogene is also a result of tyrosine phosphorylation. No new phosphotyrosine-containingproteins, however, have been detected in v-sis transformed cells. The functional similarity between the v-sis proteins and PDGF is underscored by the fact that SSV-transformed cells contain a mitogenic activity which is neutralized by anti-PDGF antibodies (178, 190, 196). Furthermore the growth of some types of SSV-transformedcell can be inhibited by these antibodies (195). It is not clear, however, whether the v-sis proteins actually have to be secreted from the cells to be active (191), since they could theoretically interact functionally with the PDGF receptor in some intracellular compartment. Cloned c-sis sequences are fully active in transformation when expressed in appropriate vectors (197, 198). Thus there are no critical structural alterations necessary for the transforming activity of the v-sis gene. However, since PDGF treatment of normal cells does not lead to transformation, this implies that there is a difference between autogenous generation of PDGF and its exogenous application to cells.

B. TGF-a

AND THE

EGF RECEPTOR

Certain transformed cells secrete growth factors that are able to cause the transient induction of the transformed phenotype in normal cells. For instance TGF-a and TGF-P in combination allow the growth of normal fibroblasts as colonies in agar. The sequence of TGF-a shows homology to EGF (199) and it has been shown that TGF-a interacts with the EGF receptor causing the same responses as EGF (200-202). Thus TGF-a stimulates the protein-tyrosine kinase activity of the EGF receptor as effectively as EGF, both in v i m and in intact cells. Since no other receptor has been identified for TGF-a, we can assume the biological effects of TGF-a are achieved by stimulating EGF receptor-mediated tyrosine phosphorylation. In this context it is worth noting that a number of tumors have been isolated in which there is aberrant expression of the EGF receptor gene (203), commonly resulting in high levels of EGF receptor expression. Often the receptor gene is amplified and in a few cases there are alterations in gene structure, some of which may lead to a truncation similar to that giving rise to the v-erb-B protein. Whether these alterations in EGF receptor expression are in any way instrumental in the tumor phenotype is unclear.

C. THEneu ONCOGENE The neu oncogene was first identified by transfection of DNA from a series of ethylnitrosourea-induced rat neuroglioblastomas onto NIH3T3 cells (204). The product of the neu oncogene was identified as a 185-kDa surface glycoprotein (pl 8 P e U )through the use of antitumor serum (205). Subsequently monoclonal

216


antibodies specific for pl 85"'" were isolated (206).p 185"'" is phosphorylated on serine and threonine and displays tyrosine-specific autophosphorylating activity in immunoprecipitates (207). Hybridization studies showed the neu gene to be related to the v-erb-B gene (208) and this enabled the isolation of molecular clones of the neu gene (209).The nucleotide sequence of the neu gene shows that it has a protein kinase domain and in this region the neu gene is closely related to the c-erb-B gene (209). The neu gene, however, is clearly distinct from the c-erb-B gene. The neu gene is very likely to be the rat equivalent of the human c-erb-B-2 gene (210). The size and properties of ~ 1 8 5 " 'strongly ~ suggest that it is a growth factor receptor. A protein very similar to ~ 1 8 5 " ~ can " be detected on normal fibroblasts (208).So far, however, the true ligand for this normal cell protein has not been identified. Treatment of neu-transformed cells with monoclonal antibodies directed against ~185"""reverses the transformed phenotype (211 ) . This reinforces the notion that ~185"'" mimics a surface growth factor receptor. It seems likely that ~185"'" is a constitutively activated form of a normal growth factor receptor which acts through tyrosine phosphorylation. D. pp60c-srcAND POLYOMA VIRUSMIDDLET ANTIGEN The initial report of a protein-tyrosine kinase activity was of that found in association with polyoma virus middle-sized tumor antigen (mT antigen) ( 1 ) . Of the three polyoma virus T antigens, mT antigen is the one most intimately connected with tumorigenesis, being capable of transforming cell lines to a malignant state in the absence of the other two T antigens. The mT antigen is a membrane-associatedprotein but is synthesized on soluble ribosomes. It does not span the membrane but appears to interact with membranes via a hydrophobic sequence near its C-terminus. Studies with site-directed mutants have shown that mT antigen must interact with membranes to transform. Analysis of a variety of mT antigen mutants shows that there is a good but not perfect correlation between the presence of associated protein kinase activity and an ability to transform (I, 212, 213). Despite strenous efforts, all attempts to demonstrate that mT antigen itself has enzymic activity have failed. Given the lack of primary sequence homology between mT antigen and other protein kinases perhaps this is not surprising. Measurement of the size of the protein kinase active fraction of mT antigen showed that it was considerably larger than the bulk of mT antigen, which behaved as a monomer (214). This suggested that mT antigen might be associated with a cellular protein kinase in the form of a complex. Subsequently it was shown that mT antigen is complexed with pp60c-src,and that the majority of the associated protein-tyrosine kinase activity can be accounted for by pp60"-"'"


217

(215, 216). Since pp6OC-"" is membrane-bound, and only mT antigens capable of binding to membranes are found in this complex (216), the interaction between the two proteins is likely to take place on the membrane. Only a small fraction of the pp60"-"" and mT antigen populations are found in association. Under some circumstances there is evidence that the amount of mT antigen is limiting, but what regulates the formation of complex is unknown. It is worth noting that in polyoma virus-infected RSV-transformed cells pp60v-"rc is not found associated with mT antigen (217). pp6OC-"" molecules which are associated with mT antigen have severalfold greater activity towards exogenous substrates than free pp60c-"rcmolecules (218). In some undefined way therefore the interaction with mT antigen increases the activity of pp60c-"rc.As previously mentioned the activity of pp6OC-"" seems to be regulated in the cell (65, 66),possibly by virtue of a tyrosine phosphorylation in the unique C-terminal region of the molecule (58). Bound mT antigen appears to prevent phosphorylation of this site (58,218a),and this may in part be responsible for the activation of the associated pp60c-srcmolecules. An affinity of mT antigen for the C-terminus of pp60c-"rc would explain its inability to associate with pp60v-src.Despite the activation of pp6OC-"", polyoma virus-infected and transformed cells show no increase in the overall level of phosphotyrosine in protein and contain no new phosphotyrosine-containing proteins (161).In this regard they are similar to cells that express high levels of pp60"-"" ( 6 5 ) .The substrates for the activated pp60"-src await detection.

E. p56Istra High levels of protein-tyrosine kinase activity were detected in membranes of the LSTRA mouse lyphoma cell line [originally isolated from a Moloney murine leukemia virus (Mo-MuLV)-induced thymoma] during a screen of suspension cells for this activity (219).Although membrane preparations from these cells can phosphorylate exogenous substrates, there is a major endogenous substrate of 56 kDa (219-222). This protein has a phosphorylation site contained in a tryptic peptide identical to that including Tyr-416 in pp6OV-""' (223). Partial purification of the protein-tyrosinekinase activity indicates that this protein, p56Istra,is almost certainly the protein kinase itself labeled by autophosphorylation (16). LSTRA cells are not unique among mouse thymomas in having elevated levels of membrane-associated protein-tyrosine kinase activity. Several other Mo-MuLV-induced and Radiation-MuLV-induced thymomas contain a protein of similar size and structure (222). A protein closely related to p56lStrais also found in normal thymocytes, albeit at lower levels, but is lacking from some cloned normal T cell lines (221-226). It is not detected in other types of hematopoetic cells, including B cells, nor in nonhematopoietic cells. B cells have a slightly smaller auto-

218


phosphorylating membrane-associated protein-tyrosine kinase, which is structurally distinct (225, 226). Like pp60v-src,p56IStrais modified by the addition of a myristyl group to its N-terminus (222, 227). cDNA clones corresponding to ~56'"""have been isolated and sequenced (228, 229) [the gene has been termed both zck (228) and lskT (229)l. The p56lStra sequence is closely related to that of the v-src and v-yes genes but is distinct from both, as well as every other cloned protein-tyrosine kinase gene. Apparently p56lSwais identical to the 56-kDaprotein in normal thymocytes (228,229),and the increased expression of p56Istrais due to the insertion of an Mo-MuLV LTR (long terminal repeat) adjacent to the ~56'""" gene. The role, if any, of p56lStrain the phenotype of LSTRA cells and the other thymomas remains unproven. LSTRA cells have higher levels of phosphotyrosinein protein than many T cell lines (221), but there is no true cognate control cell. The only phosphotyrosine-containing protein detectable in LSTRA cells is p56lSmaitself (222). F. THEv-mil, v-ruf,

AND

v-mos ONCOGENES

The v-mil gene is part of the Mill Hill 2 avian acute leukemia virus genome and is expressed as a gag-mil chimera, P1oOgag-mi~ (230). The v-ruf gene is the oncogene of the 3611-MuSV murine sarcoma virus and is also expressed as a gag-linked protein, P90gag-raf(231).Analysis of the mil and ruf sequences shows that these two genes were almost certainly derived from a cellular gene which is equivalent in chicken and mouse (232-234). The v-mos gene is carried by a number of strains of Moloney murine sarcoma virus (Mo-MuSV), and encodes an em-mos protein, ~37"" (235-237). The sequences of the predicted products of these genes have regions with homology to that of the catalytic domain of the protein-tyrosine kinase family (see Fig. 1). P 100gag-mi', P90gag-raf (230), and ~37"" (236) have all been assayed in immunoprecipitates for protein kinase activity. None of them show any detectable protein-tyrosine kinase activity. All three proteins, however, have associated activities which will phosphorylate serine and to a lesser extent threonine either in exogenous substrates or in autophosphorylation reactions (238-240). Since these proteins themselves are all phosphorylated on serine, it is hard to rule out the possibility that the detected activity is due to an associated cellular protein-serinekinase. On balance, however, it seems likely that these proteins really do have intrinsic protein kinase activity. Bacterially synthesized v-ruf (241) and v-mos (242) proteins have ATPase activities. Mutations in the ATP binding site of the v-mos gene abolish transforming activity (243).Cells transformed by these viruses do not display any perturbation of the normal tyrosine phosphorylation patterns (231, 244). It also seems probable that these enzymes are specific for serine rather than tyrosine. One indication of this specificity comes from an examination of the sequences in the region of these proteins which is homologous to the autophosphorylation site of


219

the protein-tyrosinekinases. In no case is there a tyrosine residue, but instead there are one or more serine residues which could be autophosphorylated (Fig. 1).

IV. General Properties of Protein-Tyrosine Kinases A.

SEQUENCE HOMOLOGIES

The preceding account of the individual protein-tyrosine kinases has emphasized the existence of a comparable 30-kDa catalytic domain in each enzyme. In a few cases this domain has been defined on the basis of isolation of discrete proteolytic fragments. In the others, however, the existence of this domain has been deduced largely from a striking sequence homology with the catalytic domain of pp60v-src.Figure 1 shows a comparison of the primary amino acid sequences in the relevant regions of all the protein-tyrosine kinases for which there are inferred sequences. This includes pp60v-src (5-3, pp60c-src (52), PgOgag-yes ( 6 3 , P70gag-fgr ( 7 3 , P140g'g-fps (80),P1 10gag-fes (81), P160gag-ub' (110, I l l ) , P68gag-ros (136), g ~ 6 6 / 6 8 " - ~ '(144), ~ - ~ the EGF receptor (159), gP180g'g-fms (168), and the insulin receptor (245, 246). These are compared to the sequences of four bona fide protein-serine kinases: CAMP-dependent protein kinase ( 2 4 3 , cGMP-dependent protein kinase (248), the y-subunit of phosphorylase kinase (249), and myosin light chain kinase (250). In addition the sequences of the putative viral protein-serine kinases P 1OOgag-mi' (232), F90gag-ruf (233), and ~37"" (235) are presented. The sequences have been aligned to optimize homologies. (For another perspective of sequence homologies of protein-tyrosine kinases see Chapter 7 by White and Kahn.) There are 16 residues within this domain that are absolutely conserved in all the protein kinases together with several other amino acids that are found in the great majority of protein kinases (Fig. 1). There are clusters of such highly conserved amino acids towards the N-terminus of the domain and in the Cterminal half. There is reason to believe that these may represent subdomains corresponding to an ATP binding site and a catalytic site respectively (see Section IV, B). Other regions of the domain are more variable and in some cases there are insertions of several amino acids. In addition there is the very large insertion found in the fms proteins. The homology among the protein-tyrosine kinases is greater than their homology with the group of protein-serine kinases. The converse is also true. Some comments on the specific features of these two types of protein kinase are made in Section IV, B. There is another region of sequence homology between some of the proteintyrosine kinases lying upstream of the catalytic domain. Residues 144 to 190 of pp60v-src are clearly related to regions upstream of the catalytic domain in P 9 0 g a g - y e s (residues 428-474), P70gag-fgr (residues 281-327), P140e'g-fps (resi-

220


dues 8 18-858), P85g'g-fes (residues 594-634), and P160gug-ub'(residues 244288). This region is absent in P68gUg-'OS, g~66/68'-"~-~, gP18W'g-fms, and the insulin receptor among the protein-tyrosine kinases. It is also not found in any of the protein-serine kinases nor in p37"OS, P1OOg'g-mi', and P90gUg-WThe function of this region is not yet defined although mutations in this region of pp6OV-"" affect its transforming activity (34-38), and it is necessary for P140e'g-fps (107) and P160g'g-ub' (134) to transform. The striking sequence similarities among the protein kinases imply that there was a single ancestral catalytic domain which diverged to give rise to the two classes of protein kinase specific for serine and tyrosine. It is interesting to note, however, that even among the most closely related protein-tyrosine kinases, the intron-exon structure has not been conserved. B. CATALYTIC DOMAIN

The precise limits of a competent protein-tyrosine kinase catalytic domain have not been defined, and could vary slightly from enzyme to enzyme. On the Cterminal side the leucine corresponding to Leu-516 in pp60v-srcappears to be critical. All the protein-tyrosine kinases have a hydrophobic residue at this site (either Leu or Phe) (Fig. 1). There is no correspondinghydrophobic residue in the protein-serine kinases. A two-amino acid deletion of residues 502-504, as well as all substitutions of the C-terminus of pp60v-srcwhich include Leu-5 16 abolish both protein kinase and transforming activities (34,251).Substitution of the nine residues from Pro-518 with nine amino acids from SV40 had no effect on transforming activity (252).In addition, although the sequences of pp6OV-"" and pp60c-Srcdiverge at residue 5 14, both proteins have a leucine at position 5 16. The removal of sequences beyond the residue equivalent to Leu-5 16 in gP180g"g-f"" (182) and ~37""" (253) also abolishes transforming activity. The effects of antibodies directed against specific peptide sequences provide another way of deducing regions important for catalytic activity. As predicted, antibodies against the v-src peptide 521-526 (18)or the c-src peptide 527-533 (216)are permissive for protein kinase activity, while antibodies to the src peptide 498-512 inhibit (17). Antibodies against the v-mos peptide 362-374 which ends adjacent to the leucine equivalent to Leu-516 inhibit the p37""s-associated protein-serine kinase activity (254). If Leu-5 16 is the terminal residue of the catalytic domain this puts the end of the domain very close to the C-terminus of many of the protein-tyrosine kinases (e.g., 10 residues away for pp60v-Src;17 for pp60c-src; 12 for P90g'g-yes; 10 for P70gag-fgr; 8 for P140gUg-fp"and P1 lW'g-fes; 34 for gP180eug-fms).There may be a requirement for a small C-terminal extension, but from the positioning of the catalytic domain in the EGF receptor and P160g'g-ub' it is clear there is no necessity for a C-terminal location within the protein. The boundary on the N-terminal side is less clear-cut. Homology with some of


22 1

the other protein-tyrosine kinases is evident from position 260 onwards in pp60v-”rc’(Fig. 1). The fact that the N-terminus of the y-subunit of phosphorylase kinase lies at a position equivalent to residue 249 in pp60v-srcsuggests that the catalytic domain in the protein-tyrosine kinases probably does not extend much beyond this point. A deletion extending to position 264 greatly reduces the protein-tyrosine kinase activity of pp60v-src(37), but on the other hand an Nterminal deletion mutant of ~ 3 7 in ~ which ” the N-terminus of the truncated mos protein lies at a position equivalent to 270 in pp60v-”rccan transform (253). One critical region of the catalytic domain has been identified by labeling with the ATP affinity analog, p-fluorosulfonylbenzoyladenosine(FSBA). In the CAMP-dependent protein kinase this reagent reacts with Lys-72 (255), while in pp6OV-”“it reacts with the equivalent Lys-295 (256),and in the EGFreceptor with Lys-721 (257). In all cases the derivatization inhibits protein kinase activity. A lysine in this position is absolutely conserved in all the protein kinases, being found in the sequence X-Ala-X-Lys, where both X residues are nonpolar. Based on its reaction with FSBA it seems likely that this lysine is in the vicinity of either the p or y phosphate of the substrate ATP. Site-directed mutagenesis of this position has been achieved for pp60v-”rc(258,259),~37””“(243), and P140Rag-fpS (260). Any substitution, including arginine, completely abolishes both protein kinase and transforming activity in all three cases. Some specific attribute of the 6amino group of lysine in this position must be critical. For instance if the €-amino group were simply needed to neutralize a phosphate negative charge, one would have anticipated that the guanido group of arginine would be able to substitute. Since this is not the case it seems possible that this €-amino group is directly involved in phosphorylation, perhaps mediating proton transfer. Antibodies to synthetic peptides corresponding to this region in the EGF receptor do inhibit protein kinase activity. The other residues that create the ATP binding site have not been so clearly defined. About 20 residues to the N-terminal side of this lysine, there is a cluster of glycines in the sequence Gly-X-Gly-X-X-Glywhich is also highly conserved. The only protein kinase lacking even one of these glycines is the y-subunit of phosphorylase kinase. Such a glycine-rich region is a common feature of nucleotide binding sites in many types of protein including ATPases, GTPases, and dehydrogenases. In the proteins of this type for which three-dimensional structures are available the Gly-X-Gly-X-X-Gly sequence forms an elbow around the nucleotide, with the first glycine making contact with the ribose ring and the second lying close to the terminal pyrophosphate (261-263). Mutation of these glycines has so far not been reported. A similar glycine-rich stretch is found, in several GTP binding proteins with GTPase activity, such as the transducin a-subunit, E m u , andp21““ (residues 10-15), but there is no following lysine (264,265).There are three additional regions of primary sequence homology between all these proteins which have been implicated in GTP binding based on the tertiary structure of E m u

222


(for ~21"" these are residues 58-73, 111-120, and 148-155). There are no obvious homologues of these sequences in the protein kinases. Nevertheless we can guess that some of the highly conserved residues in the C-terminal half of the catalytic domain of the protein kinases will interact with ATP, and participate in phosphate transfer. In this regard the conserved acidic residues (4 out of the 11 absolutely conserved residues in this region are acidic) seem likely to play an important role (263).The sequence Arg-Asp-Leu (residues 385-387 in pp60v-src)defines the N-terminal end of the region with the greatest homology among the protein kinases (Fig. 1). All of the protein kinases contain aspartic acid at the position equivalent to Asp-386 in pp6OV-"", and have an Asp-Phe-Gly sequence corresponding to residues 404-406 in pp60v-"rc.To our knowledge neither of these sequences has been mutagenized. The sequence Ala-Pro-Glu (residues 430-432 in pp60v-src)is found in all protein kinases except g ~ 6 6 / 6 8 ' - " ~and - ~ the EGF receptor, which have Ala-Leu-Glu, and the myosin light chain kinase, which has Ser-Pro-Glu. This sequence has been mutagenized in pp6OV-"" by substitution of individual residues at the three sites (264, 265). All the replacements tested so far abolish both protein kinase and transforming activities, but there has not been a systematic survey. It seems probable that the glutamic acid will be essential, but that there will be some flexibility at the other two sites. To the C-terminal side of this region (i.e., from residue 485 in pp60v-src)there are sequencesthat are highly conserved among the protein-tyrosinekinases, which for the most part are not represented in the protein-serine kinases (Fig. 1). It is possible that this region is involved in part in determining substrateand amino acid specificity, but there are clearly other regions of the catalytic domain involved in substrate recognition. For instance an affinity peptide substrate analog, which has been devised for the CAMP-dependent protein kinase, reacts with Cys-198 (266), a cysteine which is conserved in three of the protein-serine kinases. No equivalent cysteine is found in the protein-tyrosinekinases, but Cys-198 lies next to a glycine (residue 42 1 in pp6OV-"") which is conserved in all the protein kinases. This region is of interest because it is adjacent to the autophosphorylationsites in the proteintyrosine kinases and in the CAMP-dependent protein kinase. The autophosphorylation site in pp60v-"rc is Tyr-416. The equivalent tyrosine is also autophosphorylated in P90gag-yes, P70gag-fgr, P14Wagfps, P1 10gag-fe", and P160ga8-abl, but is not detectably phosphorylated in g ~ 6 6 / 6 8 ' - " ~ -the ~ , EGF receptor, or P68gag-r0s. Thr- 196 in the CAMP-dependent protein kinase is autophosphorylated (247).On the basis of the affinity peptide labeling studies we can deduce that this region lies very near the catalytic center. Possibly in the absence of an exogenous substrate this region forms a loop which can fold over into the active center and receive phosphate in a true intramolecular reaction. It is interesting to note that the sequences upstream of the autophosphorylation sites bear some


223

resemblance to those favored by the different enzymes in their exogenous substrates (see Section IV, C). It is evident that further progress in defining critical residues in the catalytic domains of protein kinases and their functions in phosphate transfer will require a knowledge of their three-dimensional structure. To date there are no X-ray crystallographic data for a protein kinase, but progress is being made with the catalytic subunit of the CAMP-dependent protein kinase. The ability to produce large quantities of several of the protein-tyrosine kinases in bacteria should also be an advantage in this context (267-271), although so far such proteins have been largely insoluble and therefore mostly inactive. The single exception is ptabl50, which correspondsto a fragment of the v-abl gene containing the catalytic domain (270). This protein has been purified to homogeneity and retains a high level of protein-tyrosine kinase activity. In contrast another bacterially expressed abl protein has protein-tyrosine kinase activity, but at only one-hundredth the level of ptabl50 (271). The mechanism of phosphate transfer by the protein-tyrosine kinases has not been investigated in detail. Rather few of them have been purified to homogeneity, which is a prerequisite for proper kinetic analysis. Highly purified preparations of pp60v-src(272-277), ptubl50 (270), and the EGF receptor (278-280) have been reported. In general there does not appear to be a phosphoenzyme intermediate, and the phosphate linked to the autophosphorylationsite in the catalytic domain is not turned over during phosphate transfer (91). A detailed kinetic analysis for the EGF receptor protein-tyrosine kinase using a synthetic peptide substrate suggests that the enzyme works via a sequentially ordered bi bi reaction where the peptide is the first substrate to bind and ADP is the last product to be released (281). Such a mechanism does not require a phosphoenzyme intermediate. The only residue we know for certain that is in the active center is the lysine lying close to the p or y phosphate of the ATP. All the protein-tyrosine kinases require Mg2+ or Mn2+ for activity, and presumably this is bound to the incoming ATP. The preference of many of the protein-tyrosine kinases for Mn2 when assayed in the partially purified state may not exist with the purified enzymes (270). It has been suggested that this is a result of the ability of Mn2 ions to inhibit phosphotyrosine-specific phosphatases in crude systems. In the case of the CAMP-dependent protein kinase there is evidence for a separate metal ion binding site. The fact that the optimum metal ion concentration for the protein-tyrosine kinases is considerably higher than their K,n for the metal ion-ATP complex suggests that the protein-tyrosine kinases may also have a separate metal ion binding site (270,275). Nothing is known about the chirality of the ATP required for the phosphate transfer by the protein-tyrosine kinases. Although many of these enzymes have strict requirement for ATP (e.g., PI 10gag-fes; the insulin receptor), others will use GTP instead of ATP (e.g., +

+

224


pp60v-src;ptabl50), but the K,,,for GTP (circa 100 pl4)is considerablyhigher than for ATP (10-30 pl4) in the cases where it has been measured (49, 270). . Initial experiments suggested that the turnover numbers of the protein-tyrosine kinases might be rather low compared with those of the protein-serine kinases casting some doubt on the authenticity of this phosphotransferase activity. It appears, however, that the low turnover numbers were due to a combination of inappropriate assay conditions and partial inactivation of the enzymes, which are hard to purify because of their scarcity and lipophilic nature. With the availability of better substrates in the form of synthetic tyrosine-containingpeptides and larger quantities of purified enzymes, the activities prove to be comparableto those of the protein-serine kinases. For instance pt50abr has a turnover number of 170 pmol/min/pmol using angiotensin as a substrate (270), while that for the EGF receptor is 7-55 pmol/min/pmol using synthetic peptide substrates (280-282). These values compare with a V,, of 150-200 prnol/min/pmol for the CAMPdependent protein kinase using histone as a substrate (283). The protein-tyrosine kinases that have been tested will also catalyze transfer of phosphate from phosphotyrosine-containing proteins to ADP forming ATP (270, 284).From the kinetics of formation of ATP, the free energy of the phosphate ester linkage to tyrosine in proteins can be calculated to be about 10 kcal, which is essentially the same as for the P-y phosphodiesterbond of ATP (284).In the case of pub150 the K,,,for ADP for this reaction appears to be surprisinglylow (270).In the absence of peptide substrate the CAMP-dependent protein kinase has ATPase activity (285). It is not clear to what extent this is true for the protein-tyrosine kinases; the ptabl50 protein has very low ATPase activity ( liver > brain > lung = heart > skeletal muscle. The level of 39 nmol P,/min/g of kidney tissue using (P)Tyr-albumin as a substrate is in the same order of magnitude as that of the type 1 phosphorylase phosphatase in skeletal muscle (223). The enzyme seems to be unequally distributed between the particulate and soluble fractions, with the majority being found in the latter (224, 225). Many of the problems inherent in purifying the other protein phosphatases have been encountered .with the tyrosine-specific enzymes: they also exist in a variety of forms, they are usually recovered in very low yield, many are unstable even when stored at -2O"C, and most importantly, their substrates and the kinases needed to phosphorylate them are far more difficult to prepare in large quantity. Indeed, aside from the receptors already mentioned and some undefined membrane proteins, no natural substrates for these enzymes are known. As a result, proteins such as casein, tubulin, reduced and alkylated albumin, and carboxymethylated and succinylated phosphorylase have been used to measure enzyme activity (Table VIII) (223-233). These substrates are less than ideal because though they often possess multiple tyrosyl groups that can be phosphorylated, the stoichiometry of 32P incorporation is usually low (cO.1 mol PJmol of protein). Furthermore, some of the modified proteins are highly insoluble and can be used only at low concentrations unless they are dissolved in 10 N NaOH (223), where they undergo partial hydrolysis. Most ideal would be to use a synthetic peptide corresponding to one of the phosphorylation sites of a known natural substrate. A single attempt to do so utilizing a phosphotyrosyl peptide derived from pp6OSrc

35 1

8. PHOSPHOPROTEIN PHOSPHATASES

TABLE VIll SUBSTRATES OF PHOSPHOTY ROSYL-PROTEIN PHOSPHATASES Substrate

Kinase

EGF receptor-kinase IgG Histones Carboxymethylated and succinylated phosphorylase Casein

EGF receptor-kinase pp6Ov-SK EGF receptor-kinase EGF receptor-kinase

(226-229) (29, 144, 225, 230) (195. 227, 228, 231) (226)

EGF receptor-kinase PP60v-""' EGF receptor-kinase

(29, 195, 226, 232) (144, 145) (223, 232)

Reduced, alkylated bovine serum albumin Tubulin Glutamine synthetase 67K, 50K, 37K (membrane proteins) Myosin light chain pNPP

Reference

p~60"-~~~ ?

EGF receptor-kinase -

(232) (144, 145, 223, 227. 231)

failed in that it was not dephosphorylated (232). Because of these obstacles, none of these enzymes has been obtained in a homogeneous state. In all purifications described, the enzyme does not seem to behave as a single entity, but can be separated into at least three fractions. In the first purification from detergent lysates of Ehrlich ascites tumor cells, 60% of the activity flowed through a column of DEAE-Sephadex and was not further considered. The material that was retained was eluted with NaCl and then subjected to affinity chromatography on Zn2 chelated to iminodiacetic acid-agarose. Seventy-five percent of the activity emerged with 20 mM histidine (pool 1) and the remainder with 60 mM histidine (pool 2). After further purification by gel filtration chromatography, the pool 1 enzyme displayed a specific activity of 1.2 nmol P,/min/mg versus 29 for the pool 2 material using chemically modified (P)Tyrphosphorylase as a substrate (226). A similar scheme was used to purify the rabbit kidney enzyme (223). In this instance, 70% of the phosphatase did not bind to the DEAE column and elution from Zn2 -iminodiacetic acid-agarose was carried out with EDTA instead of histidine. After further ion exchange and gel filtration steps, two enzyme species were obtained with specific activities of 213 (Peak I) and 281 nmol Pi/min/mg (Peak 11) using chemically modified (P)Tyr-albumin as a substrate. Three forms of the phosphatase have also been resolved by DEAE chromatography from bovine heart (termed Y-1, Y-2, and Y-3) (29), chicken embryo fibroblasts (pTPI, pTPII, and pTPIII) (225) and chicken brain (TI, T,, and T,) (145). +

+

352

LISA M. BALLOU AND EDMOND H.FISCHER

C. PHYSICAL PROPERTIES

Little is known about the structure of phosphotyrosyl-protein phosphatases. Several of the enzymes display sizes in the M, = 35,000-40,000 range (223, 226), similar to that of the catalytic subunits of the type 1 and 2A protein phosphatases. A comparison of a-chymotryptic fragments from the tyrosyl phosphatase Peak I and the type 1 catalytic subunit was made using reverse-phase HPLC; preliminary evidence suggested that the two enzymes are not identical but will most likely prove to be homologous (234).The peptide maps of Peak I (M, = 34,000) and I1 (M,= 37,000) phosphotyrosyl-protein phosphatases have been compared by one-dimensional polyacrylamide gel electrophoresis and it appears that these two proteins are also distinct enzymes (223). Phosphatases of larger size have also been reported, including Y-2 ( M , = 65,000) (29); T, ( M , = 30,000-100,000), T, (M,= 43,000), and T, (M, = 95,000) from chicken brain (145); and pTPI (M,= 55,000), pTPII (M,= 50,000), and pTPIII (M,= 95,000) from fibroblasts (225). It is not yet known whether these species represent multisubunit complexes.

D. ENZYMIC PROPERTIES A number of (P)Tyr-protein substrates that have been used to study phosphotyrosyl-protein phosphatases are listed in Table VIII. These have usually been phosphorylated by pp60' --src or the EGF receptor-kinase. The nonprotein substrate pNPP has also been employed because of its structural similarity to phosphotyrosine, although, as previously discussed, it is also attacked by the phosphoseryl- and phosphothreonyl-proteinphosphatases. As expected, the main distinguishing characteristic of these enzymes is their near exclusive specificity toward tyrosyl versus seryl phosphate residues. The pool 1 phosphatase of Ehrlich ascites tumor cells, for instance, dephosphorylated chemically modified (P)Tyr-phosphorylase with a V,,, of 0.17 nmol P,/min/mg (K, = 0.8 CLM) but had essentially no activity toward phosphorylase a (226). The purified cytosolic Peak I and I1 phosphatases were reported to readily dephosphorylate acidic proteins such as serum albumin, casein, and myosin light chain but did not act on basic substrates such as histone or various peptides that were rapidly dephosphorylated by calf intestinal alkaline phosphatase. It was suggested that if one were to use (P)Tyr-histone as a substrate, one would measure only the activity of alkaline phosphatase, not that of the phosphotyrosylprotein phosphatase (232). Kinetic constants for the two phosphatases from rabbit kidney are listed in Table IX. The K,,,and V,,, values are in the same order of magnitude as those reported for phosphoseryl-protein phosphatases with their preferred substrates. The Peak I enzyme had a pH optimum of 7-7.5 for both pNPP and protein

353

8. PHOSPHOPROTEIN PHOSPHATASES TABLE IX

KINETIC CONSTANTSOF PEAKSI

I1 PHOSPHOTYROSYL-PROTEIN PHOSPHATASES TOWARD VARIOUSSUBSTRATES”

AND

Peak I

Peak I1

Km

vmax

Km

Vlnm

Substrate

(pM)

(nmol Pi/min/mg)

(pM)

(nmol Pi/min/mg)

(P)Tyr-bovine serum albumin (reduced & alkylated) (P)Tyr-casein (P)Tyr-myosin light chain pNPP

12.5

18,200

I .2

240

3.6

2,000 5,600

6.6 -

540

1.4

a

200

-

-

-

2,400

Data are from Refs. (223, 232).

substrates, as opposed to pH 5-5.5 for the Peak IJ material (223). Considering that acid phosphatase attacks phosphotyrosyl residues in proteins, contamination by this enzyme cannot be discounted. Several other tyrosyl-protein phosphatases have been reported to operate best around neutrality (224-226).

E. ACTIVATORS AND INHIBITORS Because there are several different forms of phosphotyrosyl-protein phosphatase and they have been assayed under a variety of conditions, it is difficult to make definite statements about their effectors. However, some general rules seem to be appropriate: 1. Orthovanadate at micromolar levels is a strong inhibitor of the phosphotyrosyl-phosphatases as opposed to the phosphoseryl enzymes (29, 223, 225, 227, 228). This effect was first observed with the membraneassociated activity of A-43 1 cells, where 50% inhibition was obtained at 1 pM V0,-3 (228). It has been suggested that this inhibitory action might be linked to the insulin-like effect of vanadate. 2. Brautigan et al. (229) found that 10 p.M Zn2+ completely inhibited the dephosphorylation of A-43 1 cell membrane proteins by the endogenous phosphatase. In other systems substantial inhibition is seen with 5-100 p M Zn2 (223-227,230,233). This property has led to the use of Zn2+ affinity columns to purify the enzyme (223, 226). Mn2 , Mg2 , Co2 , and Ca2+ at up to 1 mM are somewhat less inhibitory (145, 225, 229). 3. pNPP is inhibitory in the low millimolar range (29, 226, 227). +

-+

+

+

354

LISA M. BALLOU AND EDMOND H. FISCHER

4. Phosphotyrosine is a competitive inhibitor that gives 50% inhibition at

5.

6.

7. 8.

9.

10.

100-400 pit4 (224,226);phosphoserine and phosphothreonine, however, are not inhibitory. A-43 1 cell membrane proteins thiophosphorylated with ATPyS were attacked at one-twentieth to one-fortieth the rate observed with normally phosphorylated proteins (235). F- , PP,, Pi, and ATP are less effective inhibitors of the phosphotyrosylthan the phosphoseryl-protein phosphatases. Fluoride at 50 mM either has little effect (145, 230) or inhibits up to 45% (29, 227); high concentrations of PP, (5-30 mM) or Pi (10 mM) inhibit 70% or more (145, 226), whereas 1 mM ATP causes almost complete inhibition of the enzyme (225). Inhibitor-1 and -2 are without effect (29, 230). Tetramisole, a powerful inhibitor of alkaline phosphatases, does not affect the activity of the tyrosine protein phosphatase of Ehrlich ascites tumor cells (226). EDTA at concentrations up to 5 mM either activates or is without effect (145, 226, 229). The Y-2 enzyme of bovine heart was activated 5- to 10fold by EDTA, with half-maximum activation at 15 pM. EGTA was a poor activator, while other chelators such as desferrioxamineor hydroxylquinoline were as effective as EDTA (29). Reducing agents such as DTT are potent stimulators of some of the enzymes (223, 224) but others, such as the Peak I1 material from rabbit kidney, remain unaffected.

F. OTHERENZYMES WITH PHOSPHOTYROSYL-PROTEIN PHOSPHATASE ACTIVITY Several other enzymes that dephosphorylate phosphotyrosyl groups in proteins have been reported. In view of their low levels of activity as compared with the tyrosyl-specific enzymes, their physiological importance is questionable. Nonetheless, they are mentioned here for the sake of completeness. Alkaline phosphatases from calf intestine, bovine liver, and E. coli will dephosphorylate (P)Tyr-histone and the EGF receptor-kinase in A-43 1 cell membranes with activities approximately 6 orders of magnitude lower than those obtained with pNPP (231).The reactions were strongly inhibited by 2 mMpNPP and 20 mM EDTA but not by 50 mM F- . Similar properties have been reported for bovine kidney alkaline phosphatase using (P)Tyr-IgG as substrate (144). An acid phosphatase in membranes of the human tumor astrocytoma dephosphorylated (P)Tyr-histones and phosphotyrosine. This enzyme activity was said to be different from classical acid phosphatases in that it did not attack the usual substrates such as P-glycerophosphate and was not inhibited by 10 mM L(+)-


355

tartrate. Histone dephosphorylation was stimulated by 5 mM EDTA and poorly inhibited by 100 pA4 Zn2+ or 10 mMpNPP, phosphotyrosine, or Pi. V04-3, on the other hand, brought about a 50% inhibition at 0.5 pi14 (236). Li et al. (237) have suggested that the predominant phosphotyrosyl-proteinphosphatase activity of human prostate is due to the prostatic acid phosphatase. The enzyme, a dimer of identical M, = 50,000 subunits, dephosphorylated (P)Tyr-IgG and (P)Tyrcasein, though at very low rates (0.5 nmol P,/min/mg) because the substrate concentration (1.56 nM) was so low. Typical acid phosphatase inhibitors such as molybdate and vanadate were strongly inhibitory at micromolar concentrations as were 5 mM L(+)-tartrate, F- , and Pi. As discussed in the previous sections, all type 2 phosphatases have been said to dephosphorylate phosphotyrosylproteins at very low rates. In many cases the two activities have different properties and may very well be due to contaminating enzymes. ACKNOWLEDGMENTS We thank Carmen Westwater and Pamela Holbeck for their assistance in typing the manuscript. Our work on the phosphatases was supported by grants from the NIH (AM 07902) and the Muscular Dystrophy Association.

REFERENCES 1. Hams, D. L. (1946). JBC 165, 541-550.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

16. 17. 18. 19. 20. 21.

Feinstein, R. N., and Volk, M. E. (1949). JBC 177, 339-346. Parnas, J. K., and Ostern, P. (1935). Biochem. Z. 279, 94. Con, C. F., and Cori, G. T. (1937). Proc. SOC. Exp. B i d . Med. 36, 119- 122. Con, G. T., Colowick, S. P., and Con, C . F. (1938). JBC 123, 375-380. Con, G. T., Colowick, S. P., and Cori, C. F. (1938). JBC 123, 381-389. Kiessling, W. (1939). Biochem. Z. 302, 50. Green, A. A,, and Con, G. T. (1943). JBC 151, 21-29. Con, G. T., and Green, A. A. (1943). JBC 151, 31-38. Cori, C. F., Cori, G. T., and Green, A. A. (1943). JBC 151, 39-55. Velick, S. F., and Wicks, L. F. (1951). JBC 190, 741-751. Keller, P. J., and Con, G. T. (1953). BBA 12, 235-238. Fischer, E. H., and Krebs, E. G. (1955). JBC 216, 121-132. Krebs, E. G., Kent, A. B., and Fischer, E. H. (1958). JBC 231, 73-83. Sutherland, E. W., and Wosilait, W. D. (1955). Narure (London) 175, 169-171. Rall, T. W., Sutherland, E. W., and Wosilait, W. D. (1956). JBC 218, 483-495. Wosilait, W. D., and Sutherland, E. W. (1956). JBC 218, 469-481. Cohen, P. (1978). Curr. Top. Cell. Regul. 14, I 17-196. Lee, E. Y. C., Silberman, S. R., Ganapathi, M. K., Petrovic, S., and Paris, H. (1980). Adv. Cyclic Nucleotide Res. 13, 95- 131. Li, H.-C. (1982). Curr. Top. Cell. Regul. 21, 129-174. Merlevede, W., Vandenheede, J. R., Cons, J., and Yang, S.-D. (1984). Curr. Top. Cell. Regul. 23, 177-2151,

356


Ingebritsen, T. S . , and Cohen, P. (1983). Science 221, 331-338. Goris, J . , Camps, T.. Defreyn, G.,and Merlevede, W. (1981). FEES Len. 134, 189-193. Khatra, B. S . , and Soderling, T. R. (1983). ABB 227, 39-51. Stralfors, P., Hiraga, A., and Cohen, P. (1985). EJB 149, 295-303. Brandt, H., Killilea, S . D., and Lee, E. Y. C. (1974). BBRC 61, 548-554. Mackenzie, C. W., Bulbulian, G. J., and Bishop, J. S. (1980). BBA 614, 413-424. Li, H.-C., Hsiao, K.-J.,and Chan, W. W. S. (1978). EJB 84, 215-225. Chernoff, J., and Li, H.-C. (1983). ABB 226, 517-530. Ingebritsen, T. S., Foulkes, J. G., and Cohen, P. (1980). FEES Lett. 119, 9-15. Antoniw, J. F., Nimmo, H. G.,Yeaman, S . J., and Cohen, P. (1977). BJ 162, 423-433. Nimmo, G. A., and Cohen, P. (1978). EJB 87, 353-365. Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 255-261. Ingebritsen, T. S., Foulkes, J. G., and Cohen, P. (1983). EJB 132, 263-274. Ingebritsen, T. S., Blair, J., Guy, P., Witters, L., and Hardie, D. G. (1983). EJB 132, 275281. 36. Ingebritsen, T. S., Stewart, A. A., and Cohen, P. (1983). EJB 132, 297-307. 37. Pelech, S . , Cohen, P., Fisher, M. J., Pogson, C. I., El-Maghrabi, M. R., and Pilkis, S. J . (1984). EJB 145, 39-49. 38. Tonks, N. K., and Cohen, P. (1984). EJB 145, 65-70. 39. Alemany, S., Tung, H. Y. L., Shenolikar, S . , Pilkis, S . J., and Cohen, P. (1984). EJB 145, 51-56. 40. Tung, H. Y. L., Resink, T. J., Hemmings, B. A., Shenolikar, S., and Cohen, P. (1984). EJB 138, 635-641. 41. Goris, J., Dopere, F., Vandenheede, J. R., and Merlevede, W. (1980). FEBSLerr. 117, 117121. 42. Yang, S.-D., Vandenheede, J. R., Goris, J . , and Merlevede, W. (1980). JBC 255, I 175911767. 43. Gratecos, D., Detwiler, T. C., Hurd, S . , and Fischer, E. H. (1977). Biochemistry 16, 48124817. 44. Resink, T. J., Hemmings, B. A., Tung, H. Y. L., and Cohen, P. (1983). EJB 133,455-461. 45. Ballou, L. M., Brautigan, D. L., and Fischer, E. H. (1983). Biochemistry 22, 3393-3399. 46. Tung, H. Y. L., and Cohen, P. (1984). EJB 145, 57-64. 47. Khandelwal, R. L., Vandenheede, J. R., and Krebs, E. 0. (1976). JBC 251, 4850-4858. 48. Goris, J., Waelkens, E., Camps, T., and Merlevede, W. (1984). Adv. Enzyme Regul. 22,467484. 49. Silberman, S. R., Speth, M., Nemani, R., Ganapathi, M. K., Dombradi, V., Paris, H., and Lee, E. Y. C. (1984). JBC 259, 2913-2922. 50. DePaoli-Roach, A. A. (1984). JBC 259, 12144-12152. 51. Brautigan, D. L., Shriner, C. L., and Gruppuso, P. A. (1985). JBC 260, 4295-4302. 52. Brautigan, D. L., Picton, C., and Fischer, E. H. (1980). Biochemistry 19, 5787-5794. 53. Merlevede, W., and Riley, G. A. (1966). JBC 241, 3517-3524. 54. Merlevede, W., Goris, J., and DeBrandt, C. (1969). EJB 11, 499-502. 55. Chelala, C. A., and Torres, H. N. (1970). BBA 198, 504-513. 56. Vandenheede, J. R., Yang, S.-D., Goris, J., and Merlevede, W. (1980). JBC 255, 1176811774. 57. Cohen, P., Yellowlees, D., Aitken, A., Donella-Deana, A., Hemmings, B. A., and Parker, P. J. (1982). EJB 124, 21-35. 58. Tellez de Inon, M. T., and Torres, H. N. (1973). BBA 297, 399-412. 59. Goris, J., Defreyn, G.,and Merlevede, W. (1979). FEBS Lett. 99, 279-282. 60. Yang, S.-D., Vandenheede, J. R., Goris, J., and Merlevede, W. (1980). FEBSLeft. 111,201204. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.


357

61. Yang, S.-D., Vandenheede, J. R., and Merlevede, W. (1981). JBC 256, 10231-10234. 62. Villa-Moruzzi, E., Ballou, L. M., and Fischer, E. H. (1984). JBC 259, 5857-5863. 63. Hemmings, B. A., Yellowlees, D., Kemohan, J. C., and Cohen, P. (1981). EJB 119, 443451. 64. Woodgett, J. R., and Cohen, P. (1984). BBA 788, 339-347. 65. Hemmings, B. A., Resink, T. J., and Cohen, P. (1982). FEBS Lerr. 150, 319-324. 66. Jurgensen, S . , Shacter, E., Huang, C. Y., Chock, P. B., Yang, S.-D., Vandenheede, J. R., and Merlevede, W. (1984). JBC 259, 5864-5870. 67. Li, H.-C., Price, D. J., and Tabarini, D. (1985). JBC 260, 6416-6426. 68. Brautigan, D. L., Ballou, L. M., and Fischer, E. H. (1982). Biochemistry 21, 1977-1982. 69. Vandenheede, J. R., Yang, S.-D., and Merlevede, W. (1981). JBC 256, 5894-5900. 70. Khatra, B. S. (1984). Proc. Soc. Exp. Biol. Med. 177, 33-41. 71. Erdodi, F., Csortos, C., Bot, G., and Gergely, P. (1985). BBA 827, 23-29. 72. Stewart, A. A., Hemmings, B. A., Cohen, P., Goris, J., and Merlevede, W. (1981). EJB 115, 197-205. 73. Ganapathi, M. K., Silberman, S . R., Pans, H., and Lee, E. Y. C. (1981). JBC 256, 32133217. 74. Foulkes, J. G., Strada, S.J., Henderson, P. J. F., and Cohen, P. (1983). EJB 132, 309-313. 75. Brandt, H., Lee, E. Y. C., and Killilea, S. D. (1975). BBRC 63, 950-956. 76. Huang, F. L., and Glinsmann, W. H. (1975). PNAS 72, 3004-3008. 77. Huang, F. L., and Glinsmann, W. H. (1976). EJB 70, 419-426. 78. Huang, F. L., and Glinsmann, W. H. (1976). FEES Lerr. 62, 326-329. 79. Nimmo, G. A,, and Cohen, P. (1978). EJB 87, 341-351. 80. Foulkes, J. G., Emst, V., and Levin, D. H. (1983). JBC 258, 1439-1443. 81. Aitken, A,, Bilham, T., and Cohen, P. (1982). EJB 126, 235-246. 82. Hemmings, H. C., Naim, A. C., and Greengard, P. (1984). JBC 259, 14491-14497. 83. Nemenoff, R. A,, Blackshear, P. J., and Avruch, J. (1983). JBC 258, 9437-9443. 84. Foulkes, J. G., and Cohen, P. (1979). EJB 97, 251-256. 85. Foulkes, J. G., Jefferson, L. S . , and Cohen, P. (1980). FEBS Leu. 112, 21-24. 86. Foulkes, J. G . , Cohen, P., Strada, S . J., Everson, W. V., and Jefferson, L. S. (1982). JBC 257, 12493-12496. 87. Khatra, B. S . , Chiasson, J.-L., Shikama, H., Exton, J. H., and Soderling, T. R. (1980). FEBS Lea. 114, 253-256. 88. Chang, L. Y., and Huang, L. C. (1980). Acru Endocrinol. (Copenhugen) 95, 427-432. 89. Aitken, A., and Cohen, P. (1982). FEBS Leu. 147, 54-58. 90. Goris, J., Defreyn, G., Vandenheede, J. R., and Merlevede. W. (1978). EJB 91, 457-464. 91. Yang, S.-D., Vandenheede, J. R., and Merlevede, W. (1981). FEBS Lerr. 132, 293-295. 92. Foulkes, J. G., and Cohen, P. (1980). EJB 105, 195-203. 93. Yang, S.-D., Vandenheede, 1. R., and Merlevede, W. (1983). BBRC 113, 439-445. 94. Gruppuso, P. A,, Johnson, G. L., Constantinides, M., and Brautigan, D. L. (1985). JBC 260, 4288-4294. 95. Aitken, A., Holmes, C. F. B., Campbell, D. G., Resink, T. J., Cohen, P., h u n g , C. T. W., and Williams, D. H. (1984). BBA 790, 288-291. 96. Defreyn, G . , Goris, J., and Merlevede, W. (1977). FEBS Letr. 79, 125-128. 97. Waelkens, E., Goris, J.. and Merlevede, W. (1984). Biochem. SOC. Trans. 12, 827. 98. Goris, J . , Waelkens, E., and Merlevede, W. (1983). BBRC 116, 349-354. 99. Goris, J., Parker, P. J., Waelkens, E., and Merlevede, W. (1984). BBRC 120, 405-410. 100. Cori, G . T., and Con, C. F. (1945). JBC 158, 321-332. 101. Yang, S.-D., Vandenheede, J. R., and Merlevede, W. (1981). FEBS Lerr. 126, 57-60. 102. Ballou, L. M., Villa-Moruzzi, E., and Fischer, E. H. (1985). Curr. Top. Cell. Regul. 27, 183192.

358


103. Ballou, L. M., Villa-Moruzzi, E., McNall, S. J . , Scott, J. D., Blumenthal, D. K., Krebs, E. G., and Fischer, E. H. (1985). Adv. Prorein Phospharases 1, 21-37. 104. Shimazu, T., Tokutake, S.,and Usami, M. (1978). JBC 253, 7376-7382. 105. Usami, M., Matsushita, H., and Shimazu, T. (1980). JBC 255, 1928-1933. 106. Tung, H. Y. L., Pelech, S., Fisher, M. J., Pogson, C. I., and Cohen, P. (1985). EJB 149, 305-313. 107. Pelech, S., and Cohen, P. (1985). EJB 148, 245-251. 108. Gergely, P., Erdodi, F., and Bot, G. (1984). FEBS Leu. 169, 45-48. 109. Hemmings, H. C., Williams, K. R., Konigsberg, W. H., and Greengard, P. (1984). JBC 259, 14486-14490. 1 10. Hemmings, H. C., Greengard, P., Tung, H. Y, L., and Cohen, P. (1984). Narure (London) 310, 503-508. 1 11. Gergely, P., and Bot, G. (1977). FEBS Lett. 82, 269-272. 112. Jurgensen, S. R . , Chock, P. B., Taylor, S., Vandenheede, J. R., and Merlevede, W. (1985). FP 44, 1052 (Abstr. 3750). 113. Tung, H. Y. L., Alemany, S., and Cohen, P. (1985). EJB 148, 253-263. 114. Tamura, S . , and Tsuiki, S. (1980). EJB 111, 217-224. 115. Imaoka, T., Imazu, M., Usui, H., Kinohara, N., ahd Takeda, M. (1983). JBC 258, 15261535. ' 116. Lee, E. Y. C., Mellgren, R. L., Killilea, S. D . , and Aylward, J. H. (1978). FEBSSymp. 42, 327-346. 117. Li, H.-C. (1979). EJB 102, 363-374. 118. Tamura, S., Kikuchi, H., Kikuchi, K., Hiraga, A,, and Tsuiki, S. (1980). EJB 104,347-355. 119. Pato, M. D., and Adelstein, R. S. (1980). JBC 255, 6535-6538. 120. Pato, M. D., and Kerc, E. (1984). Biophys. J. 45, 354a. 121. Onishi, H., Umeda, J., Uchiwa, H., and Watanabe, S. (1982). J. Biochem. (Tokyo)91,265271. 122. Werth, D. K., Haeberle, J. R., and Hathaway, D. R. (1982). JBC 257, 7306-7309. 123. DiSalvo, J., Waelkens, E., Gifford, D., Goris, J., and Merlevede, W. (1983). BBRC 117, 493-500. 124. Yang, S.-D., Vandenheede, J. R., and Merlevede, W. (1984). BBRC 118, 923-928. 125. Pato, M. D., and Adelstein, R. S. (1983). JBC 258, 7047-7054. 126. Speth, M., Alejandro, R., and Lee, E. Y. C. (1984). JBC 259, 3475-3481. 127. Pato, M. D., Adelstein, R. S., Crouch, D., Safer, B., Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 283-287. 128. Crouch, D., and Safer, B. (1980). JBC 255, 7918-7924. 129. Li, H.-C. (1981). Cold Spring Harbor Conf. Cell Proliferation 8, 441-457. 130. Usui, H., Kinohara, N., Yoshikawa, K., Imazu, M., Imaoka, T., and Takeda, M. (1983).JBC 258, 10455-10463. 131. Pans, H., Ganapathi, M. K., Silberman, S. R., Aylward, J. H., and Lee, E. Y. C. (1984). JBC 259, 7510-7518. 132. Khandelwal, R. L., Zinman, S. M., and Ng, T. T. S. (1980). BBA 626, 486-493. 133. I m a m M., Imaoka, T., Usui, H., and Takeda. M. (1978). BBRC 84, 777-785. 134. Kobayashi, M., Kato, K., and Sato, S. (1975). BBA 377, 343-355. 135. Imazu, M., Imaoka, T., Usui, H., Kinohara, N., andTakeda, M. (1981). J. Biochem. (Tokyo) 90, 851-862. 136. Killilea, S. D., Mellgren, R. L., Aylward, J. H., Metieh, M. E., and Lee, E. Y. C. (1979). ABB 193, 130-139. 137. Imaoka, T., Imazu, M., Ishida, N., and Takeda, M. (1978). BBA 523, 109-120. 138. Li, H.-C., and Hsiao, K.-J. (1977). EJB 77, 383-391.


359

Sellers, J . R.,and Pato, M. D. (1984). JBC 259, 7740-7746. Shacter-Noiman, E., and Chock, P. B. (1983). JBC 258, 4214-4219. Titanji, V. P. K., Zetterqvist, O., and Engstrom, L. (1980). FEBS Lett. 111, 209-213. Imaoka, T., Imazu, M., Usui, H., Kinohara, N., and Takeda, M. (1980). BBA 612, 73-84. Ganapathi, M. K., and Lee, E. Y. C. (1984). ABB 233, 19-31. Chemoff, J., Li, H.-C., Cheng, Y . 4 . E., and Chen, L. B. (1983). JBC 258, 7852-7857. Foulkes, J. G., Erikson, E., and Erikson, R. L. (1983). JBC 258, 431-438. Li, H.-C., Hsiao, K.-I., and Sampathkumar, S. (1979). JBC 254, 3368-3374. Schlender, K. K., and Mellgren, R. L. (1984). Proc. Soc. Exp. Biol. Med. 177, 1723. 148. Li, H.-C., and Chan, W. W. S. (1981). ABB 207, 270-281. 149. Tabarini, D., and Li, H.-C. (1980). BBRC 95, 1192-1 199. 150. Mellgren, R . L., and Schlender, K. K. (1983). BBRC 117, 501-508. 151. Kato. K., Kobayashi, M., and Sato, S . (1975). J. Biochem. (Tokyo) 77, 811-815. 152. Khatra, B. S., and Soderling, T. R. (1978). BBRC 85, 647-654. 153. Khandelwal, R. L., and Kamani, S. A. S. (1980). BBA 613, 95-105. 154. Hsiao, K.-J., Sandberg, A. R., and Li, H.-C. (1978). JBC 253, 6901-6907. 155. Yan, S. C. B., and Graves, D. J. (1982). Mol. Cell. Biochem. 42, 21-29. 156. Burchell, A., and Cohen, P. (1978). Biochem. Soc. Trans. 6, 220-222. 157. Wilson, S. E., Mellgren, R. L., and Schlender, K. K. (1983). BBRC 116, 581-586. 158. DiSalvo, J., Gifford, D., and Kokkinakis, A. (1984). Proc. SOC.Exp. Biol. Med. 177,24-32. 159. Wilson, S . E., Mellgren, R. L., and Schlender, K. K. (1982). FEBS Leu. 146, 331-334. 160. Mellgren, R. L., Wilson, S. E., and Schlender, K. K. (1984). FEBS Lett. 167, 291-294. 161. Huang, L. C., and Chang, L. Y. (1980). BBA 613, 106-115. 162. Speth, M., and Lee, E. Y. C. (1984). JBC 259, 4027-4030. 163. Wang, J. H., and Desai, R. (1976). BBRC 72, 926-932. 164. Wang, J. H., and Desai, R. (1977). JBC 252, 4175-4184. 165. Klee, C. B., and Krinks, M. H. (1978). Biochemistry 17, 120-126. 166. Klee, C. B., Crouch, T. H., and Krinks, M. H. (1979). PNAS 76, 6270-6273. 167. Cheung, W. Y., Lynch, T. J., and Wallace, R. W. (1978). Adv. Cyclic Nucleoride Res. 9, 233-25 1. 168. Richman, P. G., and Klee, C. B. (1978).JBC 253, 6323-6326. 169. Sharma, R. K., Desai, R., Waisman, D. M., and Wang, J. H. (1979).JBC 254,4276-4282. 170. Aitken, A,, Klee, C. B., and Cohen, P. (1984). EJB 139, 663-671. 171. Stewart, A. A., Ingebritsen, T. S . , Manalan, A., Klee, C. B., and Cohen, P. (1982). FEBS Lett. 137, 80-84. 172. Stewart, A. A., Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 289-295. 173. Tonks, N. K., and Cohen, P. (1983). BBA 747, 191-193. 174. Wallace, R. W., Tallant, E. A,, and Cheung, W. Y. (1980). Biochemisfry 19, 1831-1837. 175. Manalan, A. S., Krinks, M. H., and Klee, C. B. (1984).Proc. Soc. Exp. Biol. Med. 177, 1216. 176. Tallant, E. A,, and Cheung, W. Y. (1983). Biochemistry 22, 3630-3635. 177. Wood, J . G., Wallace, R. W., Whitaker, J . N., and Cheung, W. Y. (1980). J. Cell Biol. 84, 66-76. 178. Krinks, M. H., Haiech, I . , Rhoads, A,, and Klee, C. B. (1984). Adv. Cyclic Nucleoride Protein Phosphorylation Res. 16, 31-47. 179. Wallace, R. W., Lynch, T. J., Tallant, E. A , , and Cheung, W. Y. (1979).JBC 254,377-382. 180. Wolf, H., and Hofmann, F. (1980). PNAS 77, 5852-5855. 181. Klee, C. B., Krinks, M. H., Manalan, A. S., Cohen, P., and Stewart, A. A. (1983). “Methods in Enzymology,” Vol. 102, pp. 227-244. 139. 140. 141. 142. 143. 144. 145. 146. 147.

360


182. Tallant, E. A., Wallace, R. W., and Cheung, W. Y. (1983). “Methods in Enzymology,” Vol. 102, pp. 244-256. 183. Winkler, M. A,, Merat, D. L., Tallant, E. A., Hawkins, S., and Cheung, W. Y. (1984). PNAS 81, 3054-3058. 184. Merat, D. L., Hu, Z. Y., Carter, T. E., and Cheung, W. Y. (1984). BBRC 122, 1389-1396. 185. Tallant, E. A,, and Cheung, W. Y. (1984). Biochemisrry 23, 973-979. 186. Aitken, A., Cohen, P., Santikam, S., Williams, D. H., Calder, A. G., Smith, A., and Klee, C. B. (1982). FEBSLerr. 150, 314-318. 187. Kretsinger, R. H., and Nockolds, C. E. (1973). JBC 248, 3313-3326. 188. King, M. M., and Huang, C. Y. (1984). JBC 259, 8847-8856. 189. Li, H.-C., and Chan, W. W. S. (1984). EJB 144, 447-452. 190. King, M. M., Huang, C. Y., Chock, P. B., Nairn, A. C., Hemming, H. C., Chan, K.-F. J., and Greengard, P. (1984). JBC 259, 8080-8083. 191. Wolff, D. J., and Sved, D. W. (1985). JBC 260, 4195-4202. 192. Greengard, P., and Chan, K.-F. J. (1983). FP 42, 2048 (Abstr. 1701). 193. Walaas, S. I., Aswad, D. W., and Greengard, P. (1983). Nature (London) 301, 69-71. 194. Aswad, D. W., and Greengard, P. (1981). JBC 256, 348773493, 195. Chernoff, J., Sells, M. A,, and Li, H. C. (1984). BBRC 121, 141-148. 196. Pallen, C. J., and Wang, J. H. (1983). JBC 258, 8550-8553. 197. Li, H.-C. (1984). JBC 259, 8801-8807. 198. Manalan, A. S., and Klee, C. B. (1983). PNAS 80, 4291-4295. 199. Gupta, R. C., Khandelwal, R. L., and Sulakhe, P. V. (1984). FEES Lett. 169, 251-255. 200. King, M. M., and Huang, C. Y. (1983). BBRC 114, 955-961. 201. Pallen, C. J., and Wang, J. H. (1983). Can. Fed. Biol. SOC. Proc. 26, 52 (abstr.). 202. Pallen, C. J., and Wang, J. H. (1984). JBC 259, 6134-6141. 203. Matsui, H., Pallen, C. J., Adachi, A.-M., Wang, J. H., and Lam. P. H.-Y. (1985). JBC 260, 4174-4179. 204. Binstock, J. F., and Li, H.-C. (1979). BBRC 87, 1226-1234. 205. Hiraga, A., Kikuchi, K., Tamura, S., and Tsuiki, S. (1981). EJB 119,503-510. 206. Mieskes, G., Brand, I. A., and Soling, H.-D. (1984). EJB 140, 375-383. 207. Pato, M. D., and Adelstein, R. S. (1983). JBC 258, 7055-7058. 208. Li, H.-C., Tabarini, D., Cheng, Y.-S., and Chen, L. B. (1981). FP 40, 1539 (Abstr. 5). 209. Mieskes, G., and Soling, H.-D. (1985). BJ225, 665-670. 210. Feliu, J. E., Hue, L., and Hers, H.-G. (1977). EJB 81, 609-617. 211. Berglund, L., Ljungstrh, O., and Engstrom, L. (1977). JBC 252, 613-619. 212. El-Maghrabi, M. R., Haston, W. S., Flockhart, D. A., Claus, T. H., and Pilkis, S.J. (1980). JBC 255, 668-675. 213. Mojena, M., and Feliu, J. E. (1983). Mol. CeN. Biochem. 51, 103-110. 214. Erickson, R. L., Collett, M. S., Erickson, E. L., and Purchio, A. F. (1979). PNAS 76,62606264. 215. Hunter, T., and Sefton, B. M. (1980). PNAS 77, 1311-1315. 216. Collett, M. S., Purchio, A. F., and Erickson, R. L. (1980). Nature (London) 285, 167-169. 217. Sefton, B. M., and Hunter, T. (1984). Adv. Cyclic Nucleotide Protein Phosphorylation Res. 18, 195-226. 218. Pike, L. J., and Krebs, E. G. (1985). In “The Receptors” (P. Cohen, ed.). Academic Press, New York (in press). 219. Carpenter, G., King, L., and Cohen, S. (1979). JBC 254, 4884-4891. 220. Ushiro, H., and Cohen, S. (1980). JBC 255, 8363-8365. 221. Sefton, B. M., Hunter, T., Beemon, K., and Eckhart, W. (1980). Cell 20, 807-816. 222. Foulkes, J. G. (1983). Curr. Top. Microbiol. Immunol. 107, 163-180.

8. PHOSPHOPROTEIN PHOSPHATASES 223. 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237.

36 1

Shriner, C. L . , and Brautigan, D. L. (1984). JBC 259, 11383- 11390. Martensen, T. M . (1982). FP 41, 443 (Abstr. 1015). Nelson, R . L., and Branton, P. E. (1984). Mol. Cell. Biol. 4, 1003-1012. Horlein, D., Gallis, B., Brautigan, D. L., and Bornstein, P. (1982). Biochemistry 21, 55775584. Swarup, G . , Speeg, K. V., Cohen, S., and Garbers, D. L. (1982). JEC 257, 7298-7301. Swamp, G . , Cohen, S., and Garbers, D. L. (1982). BBRC 107, 1104-1 109. Brautigan, D. L., Bornstein, P., and Gallis, B. (1981). JEC 256, 6519-6522. Foulkes, J . G., Howard, R. F., and Zierniecki, A. (1981). FEBS Lett. 130, 197-200. Swarup, G . , Cohen, S., and Garbers, D. L. (1981). JEC 256, 8197-8201. Sparks, J. W . , and Brautigan, D. L. (1985). JEC 260, 2042-2045. Gallis, B . , Bornstein, P., and Brautigan, D. L. (1981). PNAS 78, 6689-6693. Moberg, E. A,, and Brautigan, D. L., personal communication. Cassel, D., and Glaser, L. (1982). PNAS 79, 2231-2235. Leis, J. F., and Kaplan, N. 0. (1982). PNAS 79, 6507-651 1. Li, H.-C., Chemoff, J., Chen, L. B., and Kirschonbaurn, A. (1984). EJB 138, 45-51.


Section ll

Control of Specific Enzymes


Glycogen Phosphorylase NEIL B. MADSEN Department of Biochemistry University of Alberta Edmonton, Alberra Canada T6G 2H7

I. Introduction

.........................

anism . . . . . . . . . . . . . . . . . . . .

............................................ 111. Molecular Structure

B. Primary Structure and Comparative Studies C. Three-Dimensional Structure IV. Substrate-Directed Control of Pho

. . . . . . . . . . . .. . . . .. . . . . .

A. Phosphorylase b Kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Phosphorylase a Phosphatase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Structural Consequences of Serine-14 Phosphorylation A. Tighter Subunit Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Other Measurable Structural Effects ...................... VI. Functional Results of the Phosphorylation of Serine-14 . . . . . . . . . . . . . . . . . A. Changes in the Allosteric Constant, L. and in AMP Binding . . . . . . . . . B. Escape from Allosteric Control . . , . . C. Characteristics Remaining Unchanged VII. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

366 361 367 368 369 369 369 310 371 313 313 314 311 311 383 384 384 386 388 389 390

365 THE ENZYMES, Vol. XVll Copyright 0 1986 by Academic Press. Inc. All rights ofreproduction in any form reserved.

366

1.

NEIL B. MADSEN

Introduction

Glycogen phosphorylase from skeletal muscle was the first enzyme found to be metabolically interconvertible between two discrete forms that exhibit different modes of allosteric control ( 1 ) . The interconversion was later shown to involve the phosphorylation of a single serine residue by a specific protein kinase (2-4) and its dephosphorylation by protein phosphatase (5). The wealth of accumulated detail available for phosphorylase and for its interconverting enzymes exceeds that for the other interconvertible enzymes and amply justifies a separate chapter. The interconverting enzymes themselves are dealt with in other chapters and the historical aspects are also expanded elsewhere in this volume. Although glycogen phosphorylase has not been reviewed in this treatise since 1972 ( 6 ) , a number of other reviews are available (7-12) which deal with various aspects of the structure-function relationships of the enzyme and with its regulation. In this review, emphasis is placed on the glycogen phosphorylases from skeletal muscle because these are the only ones for which adequate structural information is available and all comments are directed to this system unless explicitly stated otherwise. This chapter begins by briefly reviewing the state of our knowledge of the enzymic characteristics of phosphorylase. This should remind us that regulation of this enzyme by phosphorylation is superimposed on the evolutionary earlier control by feedback inhibition (glucose and glucose 6-phosphate) and by the cell’s energy charge (inhibition by ATP; activation by AMP and Pi). The molecular structure of the enzyme is reviewed because it is the best delineated of any protein dealt with in this volume and knowledge of this structure has provided us with at least partial explanations of the effects of phosphorylation and of its control. Regulation of the phosphorylation state of glycogen phosphorylase has been touched on in other chapters but here it is appropriate to deal with substratedirected controls. Especially interesting in this regard is the glucose-mediated activation of phosphorylase phosphatase in the liver because here we can see that an ancient feedback control, operating via allosteric conformational changes, has been adapted to serve as one of the modern physiological mechanisms for glucose homeostasis. The structural consequences of phosphorylating serine- 14 appear to arise chiefly from the binding of the N-terminus across the subunit interface with consequent strengthening of the intersubunit interactions within the dimer. One of the functional results of this phenomenon is to change the allosteric behavior from the sequential to the concerted model, thus resolving an old controversy w 1thin a single enzyme. Phosphorylase a, by having a covalently bound activating ligand in the form of the phosphoserine, has had its allosteric equilibrium constant (ratio between inactive and active conformers) decreased by three orders of magnitude. Therefore it has “escaped” allosteric control in that it

367

9. GLYCOGEN PHOSPHORYLASE

is no longer dependent on AMP for activity nor is the inhibition by ATP or glucose-6-P of significance. In practical physiological terms, phosphorylase activity in the muscle can be turned on and off by its phosphorylation-dephosphosphorylation. Nevertheless many of the basic characteristics of the enzyme are unaltered by its phosphorylation state, including the catalytic mechanism and substrate specificity, the binding to the glycogen particle, and the inhibition by glucose binding to the active site and compounds binding to the negative effector site.

II. Enzymic Characteristics of Phosphorylase A.

BIOLOGICAL ROLE

Glycogen phosphorylase ( 1,4-a-~-glucan:orthophosphate a-D-glycosyltransferase, E.C. 2.4.1.1) catalyzes the first step in the intracellular degradation of glycogen as shown in Scheme I. (a-I,4-Glucoside),, + Pi G (a-l,4-glucoside),,I +

a-D-glucose-I-P

SCHEMEI when n represents the number of glucosyl residues. Although the measured equilibrium constant is somewhat less than one at neutral pH (13), the reaction in vivo proceeds as written because the intracellular concentration of Pi greatly exceeds that of glucose-1-P (14). Glycogen is a highly branched molecule and phosphorylase can degrade only to within four glucosyl units of an a-I ,6-linked side chain. A second enzyme, colloquially known as the debranching enzyme, is needed to deal with the branch points. It is not known to be under any control, but is has been suggested to become rate limiting in muscle when phosphorylase is fully activated (15). The role of phosphorylase, then, is to supply the glycolytic pathway with a regulated amount of phosphorylated glucose units derived from tissue glycogen stores. In muscle this function is regulated according to the energy needs of contraction, whereas in liver the enzyme’s activity is modulated to provide enough glucose to maintain the fasting blood-sugar level. In other tissues the enzyme may respond to emergency situations such as anoxia. Glycogen is distributed widely in nature and its utilization via phosphorylase appears to be related to energy needs in growth, development, or starvation, as well as for the purposes previously discussed. Electron microscopy reveals the presence of large-molecular-weight aggregates of glycogen in both muscle (16) and liver (17). These “glycogen parti-

368

NEIL B. MADSEN

cles,” when isolated under mild conditions, contain the majority of the phosphorylase in both tissues, as well as the other enzymes associated with glycogen metabolism (18, 19). Phosphorylase (and glycogen synthase) are rather unique because, while quite soluble and easily extracted in aqueous media, they are rendered particulate by binding to their substrate via separate “glycogen storage” sites. Certainly for phosphorylase, and possibly for the synthase, this mode of binding orients their “control faces” toward the cytosol and permits metabolic interconversion while still bound to the glycogen. B. KINETICS AND MECHANISM The kinetic mechanism of glycogen phosphorylates from liver, muscle, and some other tissues has been established as rapid-equilibrium random bi bi through the use of conventional kinetic studies (20-22) and, in some cases, isotope exchange at equilibrium (23, 24). The latter technique provides convincing evidence that the kinetic mechanism is not altered by the interconversion between the a and b forms, nor by changing the normal Michaelian kinetics to the allosteric type in the presence of ATP or glucose (25).This type of kinetic mechanism is useful for studying an allosteric enzyme because the apparent Michaelis constants are, in principle, equivalent to dissociation constants. It does not mean, however, that a covalently bound reaction intermediate might not occur during the catalytic process. The catalytic mechanism has proven to be very elusive but the key to it must be the presence in the active site of pyridoxal phosphate. Since the possible roles of this coenzyme in the mechanism have been reviewed (12) as well as extensively discussed in the literature (26, 27), only a cursory summary is given here. One may remember, too, that a complete understanding of catalytic mechanisms is not necessary for an elucidation of their biological regulation. The role of pyridoxal phosphate in catalysis by phosphorylase is recognized as being unconventional because years of painstaking research have eliminated consideration of all parts of the coenzyme except the phosphate moiety, and that is essential (12). Both crystallographic (28, 29) and chemical (30, 31) studies suggest that the phosphate of the coenzyme may interact with the phosphate of the substrate but the precise involvement remains unclear. Discussion revolves around the possibility that the phosphate acts as a proton donor (acid catalyst) (27, 32) or as an electrophile (26, 33, 34). Since the active site is buried deeply within the monomer, it is not surprising that conventional chemical modification studies have not been very rewarding. Although modification of a single carboxyl group causes inactivation (35), there is no evidence that this group is within the active site, although a carboxyl group is an attractive candidate for involvement in an enzyme-bound glucosyl intermediate. Lys-573 and Arg-568 are found within the active site by


369

crystallographic analysis and are also located there by chemical means (36, 37). The former may bind the phosphate of the coenzyme while the latter may interact with the phosphate substrate. Other groups within the active site that may be involved with catalysis include Asp-283, Tyr-572, and His-570. The latter may account for the alkaline pK of 7.1 in the pH-activity curve (38). Finally, several lines of research provide indirect evidence for a glucosyl-enzyme intermediate in the catalytic reaction (39). C. ALLOSTERIC CONTROL Manifestations of allosteric control were discovered very early, consisting of the requirement for AMP for the activity of phosphorylase b in 1936 (40),the inhibition by glucose in 1940 (13), and the inhibition of phosphorylase a by glucose in 1943 (41). In the latter case, Cori, Cori, and Green showed clearly that the kinetics for the substrate, glucose 1-P, change from first to second order. Phosphorylase a, of course, was shown to have no requirement for AMP although its activity was stimulated by 50% and its dissociation constant for this effect was 1.5 pM instead of 30 to 50 pM for the b form (in the presence of saturating substrates). Parmeggiani and Morgan discovered the inhibition of phosphorylase b by ATP in 1962 (42) and from then on a number of laboratories have published many papers dealing with the allosteric phenomena in this enzyme. Several reviews may be consulted for details (7, 8, 11)and further discussion of allostery are deferred until the effects of serine-14 phosphorylation are discussed in Section VI.

111.

Molecular Structure

A.

SUBUNIT RELATIONSHIPS

Glycogen phosphorylase was the first enzyme shown to exist in oligomeric form. The first step in this exposition was the discovery that the interconverting enzyme, which we now know is a protein phosphatase, not only caused a change in the enzymic characteristics when converting the a form to the b form, but also caused a concomitant halving of the molecular weight (43). Not long afterward it was discovered that titration of sulfhydryl groups with mercurials resulted in a correlated loss of activity and an additional halving of the molecular weight, so that phosphorylase a was shown to be a tetramer of identical subunits whereas phosphorylase b was a dimer (44). The initial assumption of identical subunits, a reasonable interpretation of the then available data, was questioned more than once and cannot be considered to have been proved until much later when sufficient amino acid sequence data became available. The reader should be

370

NEIL B. MADSEN

reminded that the difference in oligomeric structure between the a and b forms is a rather unique feature of some skeletal muscle systems, since the phosphorylase a enzymes of liver (45), lobster muscle (46),and rabbit heart (isozyme I) (47) are dimers. While AMP, especially in the presence of magnesium, may promote a tetrameric form of phosphorylase b (48), the functional oligomer for both forms of the muscle enzyme appears to be the dimer, since the latter has a higher affinity for glycogen and a higher specific activity (49). STRUCTURE AND COMPARATIVE STUDIES B. PRIMARY The amino acid sequence of phosphorylase was announced in 1977 GO),just in time to play a key role in the crystallographic studies. Each monomer contains 841 amino acids and together with the coenzyme and the phosphate on serine-14, yields a molecular weight of 97,412 for phosphorylase a. Since then, almost complete sequences have been obtained for the maltodextrin phosphorylase of E. coli (51) and for the phosphorylase found in potato (52). In addition, sequences surrounding the phosphorylatable site and the coenzyme site have been determined for phosphorylases from yeast (53)and dogfish muscle (54). Although these authors have commented on the significance of their findings, a thorough review and assessment is still awaited. I have reviewed these findings briefly (12) and reiterate the main points of interest in this chapter. There is almost 100% homology for the residues surrounding the coenzyme for all these phosphorylases. In addition, where the residues comprising the active site are known (rabbit muscle, E. coli, and potato enzymes) the homology also approaches 100%. It is only slightly less for residues comprising the twisted beta-sheet cores of the N-terminal and C-terminal domains but considerably less for the outer helices and loops. This suggests, as noted by Palm et al. (51), that the basic structure of all phosphorylases is remarkably similar. The greatest dissimilarities occur in the residues comprising the glycogen storage site of the rabbit muscle enzyme (see below), since the E. coli and potato enzymes have no such sites, and the N-terminal residues associated with regulation, both phosphorylation and AMP binding. Thus the E. coli enzyme lacks the first 17 residues of the muscle phosphorylase, and hence the phosphorylatable serine. There is virtually no homology until residue 80, so that the nucleoside subsite for AMP binding cannot be located while the arginine residues comprising the subsite for the phosphate moiety are present, suggesting a common primitive phosphoryl binding site which was extended in some eukaryotes. The potato phosphorylase also shows little homology until residue 80 of the muscle enzyme sequence is reached. In the dogfish enzyme, 12 of 14 residues surrounding the phosphorylatable serine are identical with those of the muscle enzyme, and the interconvertible enzymes are freely interchangeable. However, the phosphorylatable residue in yeast phosphorylase is a threonine in a sequence that is not similar


37 1

to those that contain the phosphorylatable serine in other phosphorylases. The sequence studies available present a consistent picture of an ancient enzyme that has shown very strong conservation of the sequences concerned with the catalytic mechanism and basic structure of the protein. The major evolutionary divergence has been concerned with the N-terminal residues where most of the control elements are located, including interconversion by phosphorylation, AMP binding and activation, and subunit interactions.

C. THREE-DIMENSIONAL STRUCTURE Crystallographic studies in Oxford and Edmonton (later San Francisco) have developed structures of phosphorylase b at 2.0 8, resolution (55-58) and phosphorylase a at 2.1 8, resolution (10, 59, 60) although full descriptions of the molecules at high resolution have not been published. The crystals of phosphorylase a were grown in the presence of glucose, which is found in the glucose-binding subsite of the active site, and the structure is therefore that of the inactive T conformation. This presents severe problems in trying to assess the catalytic mechanism but the general structure, features of the phosphorylatable serine site, and the ligand-binding sites can all be determined. The phosphorylase b crystals, grown in the presence of IMP, are isomorphous with those of the a form and also contain the enzyme in its T-state conformation, although one or two features of an expected active R-state are apparent. In general the structures of the a and b forms are quite similar except at the control sites on the Nterminus, as discussed in Section V. The structure of phosphorylase is typical of the OL and P class of proteins. Two major domains each consist of a core of twisted P-sheet surrounded by a-helices. The domains are associated closely together, with their interface containing the pyridoxal phosphate and the large cavity of the active site, the latter being some 15 A from the surface of the protein. The two subunits of the dimer are closely associated through contacts between the N-terminal domains and it is via this interface region that both homotropic allosteric interactions between subunits and heterotropic interactions within a subunit are transmitted between ligand-binding sites (58, 61, 62). Furthermore, the effects of serine-14 phosphorylation are manifested through changes in these intersubunit contacts, discussed in Section V. Although the two subunits of the phosphorylase dimer are related by a twofold axis of symmetry, the dimer is remarkably asymmetric, as we illustrated by computer-drawn space-filling models (62). One can define a concave ‘‘catalytic face” which is bound to the glycogen molecule by the glycogen storage sites on the periphery of this face. Closer to the diad axis is found the active site cavity. Ligand-binding and/or kinetic studies have shown that glucose or stereochemically similar compounds bind in the glucose subsite of the active site and pro-

372

NEIL B. MADSEN

mote the T conformation in both a and b forms of phosphorylase (60, 63-66). Glucose-1-P and analogues such as UDPglucose and glucose 1,2-cyclic phosphate bind in much the same location but promote the R conformation (55, 6769). On the solvent side of a protein loop that forms part of the active site is found an inhibitor site that binds purine derivatives such as inosine, adenosine, and caffeine. The effects of these compounds are synergistic with those of glucose because they and glucose stabilize each other’s binding sites. Thus Asn-284 forms a hydrogen bond with the 2-hydroxyl of glucose whereas, on the outside of the loop, the phenyl ring of Phe-285 sandwiches the base between itself and Tyr-612 (60, 70).Because the loop containing Asp-283, Asn-284, and Phe-285 must move to accommodate glucose- 1-P or phosphorylated glucose inhibitors (69),glucose and caffeine exhibit competitive inhibition with respect to glucose-1-P and are synergistic in stabilizing the T-state conformation for both a and b forms of the enzyme. On the side of the dimer opposite the catalytic face one can define a convex “control face.” Here are found the two binding sites for AMP, each consisting of a phosphoryl site of two or three arginines, while the nucleoside portion interacts with several residues including some for the symmetry-related subunit (71). Important among these residues is Tyr-75 because chemical studies indicate it is involved in AMP binding (72, 73). It does not interact with AMP in the T-state phosphorylase a (63) but it does interact with the purine base of AMP in the phosphorylase b crystals (71).This suggests a vaguely defined mechanism for the movement of helix 49 to 75 which is seen upon the addition of substrate to the phosphorylase a crystals (64,as well as a means for communicating homotropic cooperativity between AMP-binding sites since the above mentioned helix connects these. Also binding to the AMP site are the inhibitors ATP and glucosed-P, showing slightly different binding modes and causing different conformations of adjacent protein groups, as has been well defined for the physiologically significant case of phosphorylase b (58, 74). Although the binding sites for AMP and glucose-6-P do not overlap, their respective phosphates both bind to Arg-309, which occupies different positions in the two cases and explains the exclusivity of binding observed previously. Close by the AMP-binding site is the site for phosphorylation, Ser-14. The phosphate of the latter, as seen in crystalline phosphorylase a, forms a salt bridge to Arg-69 of helix 49-75, just to the solvent side of AMP, thus forming an obvious link between the two control sites. Since residues 1-18 cannot be visualized in crystalline phosphorylase b and are assumed to be flexible (75) whereas all but the first 4 are seen in phosphorylase a, one of the obvious effects of the phosphorylation of phosphorylase b is to create a covalently attached ligand which now lies across the intersubunit interface and strengthens it by the formation of new salt bridges and other interactions (76). Therefore among the ligand-binding sites on phosphorylase a may be placed the two for the serine- 14


373

phosphate (Arg-69 from the same subunit and Arg-43 from the symmetry-related subunit), and one for each of Arg-10 and Lys-1 1, which form hydrogen bonds to Gly- 1 16 and Tyr- 113, respectively, from the symmetry-related subunit. * The effects of this intersubunit binding of the N-terminal residues of phosphorylase a are discussed in Section V.

IV. Substrate-Directed Control of Phosphorylation and Dephosphorylation

A. PHOSPHORYLASE b KINASE Krebs et al. (77) showed that glycogen decreases the K,,, of phosphorylase b for nonactivated kinase by a factor of 10, and that this was not likely to be due to its stimulation of kinase activation by direct binding. Tabatabai and Graves (78) found that glycogen had no significant effect on the kinase action on the synthetic tetradecapeptide containing residues 5 to 18. We may therefore assume that when phosphorylase b is bound to glycogen, leaving the control face exposed, there are sufficient conformational changes to improve the binding of the kinase. With a rapid-equilibrium random bi bi mechanism (78), the K, should be a measure of binding. Presumably, with a mobile N-terminus, the effect of glycogen is on residues surrounding Ser-14 other than the first 20. The promoter of the allosteric R-state, AMP, does not have a significant effect on kinase activity (77), nor does another substrate, glucose-l-P. The allosteric inhibitors, glucose-6-P and ATP, do inhibit kinase activity on phosphorylase b (77). The ATP effect is difficult to analyse since there is inhibition of the kinase when acting on the synthetic peptide (78) but the inhibition by glucose-6-P is unequivocably substrate-directed because kinase action on the synthetic peptide is not affected (79). The latter study also showed that inhibition by glucose-6-P is competitive with respect to phosphorylase b, noncompetitive to MgATP, suggesting that the sugar phosphate, when bound to the activator site, promotes a conformation of phosphorylase that binds poorly to the kinase. Covalently bound AMP prevents the inhibition. A study employing cross-linking reagents suggests that AMP does not affect the motility of the N-terminus (80). Although this conclusion has not been confirmed by definitive protein chemistry, it would be consistent with the general concept that substrate-directed effects on phosphorylase kinase activity are not mediated via the N-terminus but through changes in the adjacent protein structure, which constitutes the kinase-binding

*Personal communications from Dr. R. J. Fletterick, reported also in his address before “The Robert A. Welch Foundation Conferences on Chemical Research,” XVII, Houston, Texas (1983).

374

NEIL B. MADSEN

site and which lowers the K , by 70-fold in comparison to the synthetic peptide (78).

B. PHOSPHORYLASE a PHOSPHATASE The major portion (-75%) of the phosphorylase a phosphatase activity in skeletal muscle is accounted for by protein phosphatase-1, as designated by Cohen and his colleagues (81). Furthermore, protein phosphatase- 1 is virtually the only activity found in the glycogen particle (81, 82). Therefore the preparation of phosphorylase a phosphatase utilized in the studies of Martensen et al. (83, 84) was most likely composed chiefly of protein phosphatase-1 and, as discussed in (82), that studied by Detwiler et al. (85) almost certainly was. Martensen et al. (83, 84) utilized both phosphorylase a and the phosphorylated tetradecapeptide (residues 5- 18 of phosphorylase a ) to determine whether ligands affected the phosphatase directly or via interaction with the substrate. Surprisingly, inorganic phosphate, a substrate for phosphorylase, was a competitive inhibitor for either phosphatase substrate, a result compatible with the idea that this phosphatase may have the ordered uni bi kinetic mechanism commonly found for hydrolytic enzymes. AMP, as shown earlier in (86), caused inhibition that was substrate directed, as was true also for glucose-1-P because there was no inhibition with the phosphorylated peptide as substrate. Both AMP (83, 85) and glucose- 1-P (83) caused inhibition which was superficially competitive with phosphorylase a, which was true also for UDPglucose (85). Since all these compounds promote the R-state of phosphorylase a , one may assume that the latter either binds poorly or not at all to the phosphatase or is a poor substrate. Martensen et al. (84)also studied the activation of the phosphatase by metabolites and demonstrated that glucose, glucose-6-P, and glycogen all acted by binding to the substrate phosphorylase a. The first two caused large increases in maximal velocity with no significant change iii the K,. Glycogen did not affect the V,,, but lowered the K,,, significantly. Detwiler et al. (85) examined the effects of metabolites on the kinetics of muscle phosphorylase a phosphatase acting on the natural substrate. They concluded that the AMP-phosphorylase a complex was a poor substrate with a V,,, only 5% of the unliganded substrate but with a K,,, increased by only 50%. Even for this kinetic mechanism, the results suggested that the affinity between phosphatase and substrate has not been altered significantly and that the inhibition affects primarily the velocity of the catalytic reaction. Similarly, the activation by glucose or glucose-6-P resulted in a three- or twofold increase in V,,, with little change in K,,,. Caffeine, long known to be an activator (87),was also shown to act through the substrate (86). With the exception of ADP and ATP (85), which cause inhibition, all the metabolites that stabilize the inactive T conforma-


315

tion are activators, whereas all the compounds that promote the active R configuration of phosphorylase a are inhibitors of the phosphatase. The allosteric control is therefore reinforced or superceded with a compatible control of covalent interconversion; ATP and ADP may not be an exception because they inhibit phosphorylase a only slightly, as does glucose-6-P. These compounds, while binding at the activator site, may cause local changes around the serine-14 phosphate without affecting the active site. The substrate-directed activations and inhibitions of phosphorylase phosphatase have been interpreted at the molecular level from considerations of the X-ray-derived crystal structure of phosphorylase and the effects on that structure of allosteric effectors (61, 62). We suggested that AMP, and/or the active-site ligands glucose- I-P and UDPglucose, stabilize the active R conformation of phosphorylase a to which the phosphatase may bind but in which the serine phosphate is firmly bound and unavailable to the active site of phosphatase. This presumed firm binding of the serine phosphate and adjacent peptide chain is consistent with the crystallographic studies (61), NMR data (88), cross-linking studies (89), and protection of the N-terminal chain against proteolytic attack (90). Glucose and caffeine, either singly or acting synergistically together, stabilize the inactive T conformation in which the serine phosphate is exposed to catalytic action by the phosphatase but in which the surrounding binding site for the modifying enzyme is not altered significantly. Parallel with these developments in the muscle system, Hers and Stalmans and their collaborators developed the hypothesis that phosphorylase a is the glucose receptor in the liver cell (91-93). They demonstrated in vivo, in isolated hepatocytes, and in vitru, that glucose activates phosphorylase a phosphatase and that phosphorylase a inhibits the phosphatase for glycogen synthase. In vivo, a glucose load results in the conversion of phosphorylase a to b and, after a lag period of approximately the two minutes needed to reduce the proportion of phosphorylase a to less than 10 to 20% of the total, the activation of glycogen synthase and deposition of glycogen. The molecular basis for the in vitro behavior of the muscle system may also be applied to the liver system. Even if there are separate phosphatases for phosphorylase and synthase, inhibition of the one for synthase by phosphorylase a until the concentration of phosphorylase a is reduced below a minimal value is consistent with the mechanism of glucose activation. The very poor binding of phosphorylase b to the phosphorylase phosphatase ( K , = 140 pM)is consistent if this phosphatase is the one that also acts on the synthase in response to the physiological stimulus to glucose. Work by Cohen and collaborators has shown that the collective activities of protein phosphatases-2A in rabbit liver account for 60% of the activity with phosphorylase a as substrate and 40 to 70% with the various phosphorylated sites of glycogen synthase (81).Therefore control of a single phosphatase, inhibited by phosphorylase a in the presence of AMP and substrates, activated by glucose,

376

NEIL B. MADSEN

and then released from the phosphorylase system to act on glycogen synthase and other enzymes when there is no longer sufficient phosphorylase a left to bind it, provides the simplest explanation for the observed effects of glucose. This does not preclude additional controls for the same or other protein phosphatases. Kasvinsky et al. (94) showed that caffeine increases the glucose-induced phosphatase action on phosphorylase a in intact heptocytes, but it is not known if there is a natural metabolite in liver that utilizes the inhibitor site on phosphorylase a. Monanu and Madsen (95) have studied the substrate-directed regulation of liver protein phosphatases-2A and -2A, acting on liver phosphorylase a and have confirmed that the essential elements elucidated for the muscle system are present. Both AMP and ATP are particularly potent inhibitors that require glucose and caffeine acting together for reversal. Ligands have much greater effects on maximal velocity than on K,. In assessing the physiological significance of these substrate-directedmetabolite regulations of phosphorylase a phosphatase activity, one may note, as discussed by Martensen et al. (84), that the concentration of glucose-6-P in frog sartorius muscle subjected to tetanic stimulation may rise to 3.6 mM. The concentration level and the timing suggest that this metabolite may play a role in the decline of phosphorylase a observed after prolonged stimulation. The activation of the phosphatase is complemented by the inhibition of phosphorylase kinase (79). Since little if any free glucose is observed in muscle, it can play no role in the regulation of protein phosphatase activity in that tissue. In liver, however, with an in vivo glucose concentration equivalent to that of the serum, and with the wealth of physiological evidence provided by Hers and his colleagues, we can assume that there is a major physiological role for glucose as a regulatory metabolite. The coenzyme, pyridoxal phosphate, probably plays no physiological role in the regulation of phosphorylase a phosphatase in muscle, but a comprehensive study demonstrated that its absence or alteration affected phosphatase activity and regulation significantly (96). Apophosphorylase a was dephosphorylated more rapidly than the native enzyme but glucose or caffeine did not increase the rate, whereas AMP was still able to inhibit. Examination of dephosphorylation rates of phosphorylases reconstituted with various pyridoxal phosphate analogs showed that, in general, only analogs that restore catalytic activity to the phosphorylase would restore the effects of glucose and caffeine and, in particular, this meant having two ionizable hydrogens at the 5’-position. Therefore the coenzyme is needed for the transmission of conformational changes in at least the case of caffeine, which can dissociate pyridoxal-reconstituted phosphorylase a from tetramer to dimer without increasing the phosphatase activity. This example, as well as others given, show that the activating effects of glucose and caffeine cannot be explained completely by suggesting that it is due to the dissociation of tetramers to dimers (97). Furthermore, the idea that the phos-


377

phatase dephosphorylates only the dimeric form (97) was contradicted by the finding that glucose increases by 6-fold the phosphatase action on phosphorylase a reconstituted with the phosphoethenyl analog of pyridoxal phosphate without causing it to dissociate (96). We should remember, too, that the substratedirected control of liver phosphorylase a phosphatase cannot operate via alterations in the oligomeric structure.

V. Structural Consequences of Serine-14 Phosphorylation A. TIGHTERSUBUNIT INTERACTIONS As mentioned in Section III,C, crystallographic studies on phosphorylase b failed to locate the first 18 residues of the N-terminus ( 7 3 , whereas all but the first 4 are well defined in the structure of crystalline phosphorylase a (59) where it lies across the subunit interface. These differences between the a and b forms in the crystalline state are consistent with solution studies published before and after the structures became available. For example, it was shown that subtilisin carries out a limited proteolysis of phosphorylase b, cleaving on the carboxy side of residues 16 and 264, whereas with phosphorylase a the activity is confined largely to residue 264 (50, 98). The slight activity of subtilisin on the N-terminus of the a form is decreased even further by AMP and increased only slightly by glucose or glycogen (99). The latter finding appears to rule out the tetrameric state of phosphorylase a as a reason for differences in the proteolytic action and focuses attention on the conformation of the N-terminus. The marked acceleration of subtilisin action at residue 16 by glucose-6-P (99) agrees with the effect of this ligand on phosphatase activity, as previously discussed. It is interesting to note that all ligands tested, whether promoters of the inactive T-state or of the active R-state, decreased the inactivation rate by subtilisin (99), suggesting a tighter association of the “tower loop” (containing the Gln-264-Ala-265 bond) with the symmetry-related subunit. The structure of the N-terminus and its association with the main body of the protein was described at a resolution of 2.5 A (59, 62), and the extra stabilizing effect on the stimuli interactions of the dimer has been discussed extensively (76). Figure 1 depicts interactions of the N-terminus with residues from both subunits, drawn from more highly refined coordinates at a resolution of 2.1 A. The phosphate makes on intrasubunit salt bridge to Arg-69 of the helix containing the Tyr-75 which may well bind to the adenine moiety of AMP in the active R conformation. The role of this salt bridge in communicating between the serine phosphate and the activating AMP is discussed in Section III,C (61). A new intersubunit salt bridge is formed, between the serine phosphate and Arg-43. In addition, two new hydrogen bonds are formed between Arg-10 and Lys-11 to

378

NEIL B. MADSEN

FIG. 1. Stereo diagram to illustrate the interactions of the phosphorylated N-terminus with residues from both subunits. Residues 65-70 and 5-15 (the N-terminus) are from one subunit; all others are from the symmetry-related subunit. The tetrahedral phosphate on Ser-14 (to the left of Val-15) interacts with Arg-69 from above and Arg-43’ from the other subunit (both distances = 2.6 A). An -NH2 of Arg-10 interacts with the carbonyl oxygen of Gly-116’ (2.9 A) (just below the caption for Leu-115’). The r-amino group of Lys-9 forms a similar hydrogen bond with the carbonyl oxygen of Tyr-113‘ (3.0 A) near the bottom of the diagram. (The side chain of Tyr 113’ is the lowest feature shown.) Figure I was drawn by Dr. R. J. Fletterick from coordinates of phosphorylase a at 2.1 8, resolution refined to an R-factor of 0.19 (S. Sprang and R. J. Fletterick, unpublished data).

Gly-I16 and Tyr-113, respectively, of the symmetry-related subunit, and steric complementarity appears favorable for van der Waals contacts.* The two new salt bridges are in addition to the two found in both a and b forms, the latter being inaccessible to solvent and hence stronger. The two new salt bridges, while relatively accessible to solvent, are surrounded by polar and nonpolar contacts which may increase their stability. An NMR study indicated that the phosphate is titratable, hence accessible to solvent, and mobile (at least in the glucoseliganded protein) but behaves as though interacting with positive charges (100). One may assume that these salt bridges and their supporting environment, as well as the hydrogen bonds and sterically favorable van der Waals contacts, are the chief causes of the considerably increased stability and subunit interactions displayed by the phosphorylase a dimer as compared to dimer b. Figure 2 illustrates a striking feature of the structure in the area of the N*Personal communications from Dr. R. J. Fletterick, reported also in his address before “The Robert A. Welch Foundation Conferences on Chemical Research,” XVII, Houston, Texas (1983).

FIG. 2. Stereo diagram of the surface layer of residues surrounding Ser-14-P. Residues from the symmetry-related subunit are drawn in broken lines and the residue numbers are primed. This diagram was drawn from coordinates at a resolution of 2.5 A and was published in different form in Ref. (62).

380

NEIL B. MADSEN

terminus. The negative charges of the serine phosphate are surrounded by positive charges derived from both subunits, while all but one of the nearby negative charges (Asp-32) are either oriented away from the surface or are outside the ring of positive charges. The eleven positive charges are His-73, Lys-839, Arg-69, Arg-66, Arg-16, Arg-10, Arg-33’, Lys-29’, Lys-28‘, Arg-43’, and His-36’, where primed numbers indicate the symmetry-related subunit. The possible significance of these positive charges in providing a binding site for the phosphatase has been discussed thoroughly (62) and the analogy to similar asymmetric positive charge distributions in cytochrome c was pointed out, the latter providing binding sites for the enzymes that interconvert cytochrome c between its oxidized and reduced forms. Long before we knew the three-dimensional structure adjacent to the serine phosphate, the presence of four positive charges in the contingent amino acid sequence suggestedto Fischer er al. (4)that the neutralizationof these charges by the covalently attached phosphate permitted interactions at this locus which were previously electrostatically repulsed, thus leading to the activation of the enzyme and its association into tetramers. Sealock and Graves (101) studied the effects of various salts on phosphorylase activity and concluded that the interaction of the covalently bound phosphate with groups at a specific site is very sensitive to the ionic environment. The predictions of both these groups are consistent with the available structure. Pertinent to our discussion of the location of the N-terminus is an immunological study carried out by Janski and Graves (102). They showed that antibodies specific for the N-terminal region of phosphorylase a are highly specific for only the first four amino-terminal residues of the enzyme. This agrees with the failure to locate these residues in the crystal structure. Furthermore, only one molecule of antibody binds per mole of dimer, causing a strong inhibition of both kinase and phosphatase activity but with less effect on the catalytic phosphorylase activity or K, for glucose-1-P and AMP, and a decreased K, for glycogen. Interestingly, the antibody stabilized phosphorylase b against resolution of the pyridoxal phosphate under the usual conditions, thus making it more like phosphorylase a and suggesting that the antibody is cross-linking the two monomers and preventing the dissociation necessary for resolution. As discussed with regard to subtilisin, the N-terminal region is susceptible to proteases and it has long been known that trypsin will cause the formation of a pseudophosphorylase b species termed phosphorylase b’ (103,104).This species is active in the presence of AMP but it does not exhibit homotropic cooperativity for AMP or glucose-1-P and heterotropic interactions are greatly reduced (90). It is known that phosphorylase b’ lacks the first 16 residues (4) and will bind the phosphorylated tetradecapeptide (residues 5- 18) ( 1 0 3 , with the resultant induction of enzymic properties similar to those of phosphorylase a. The same result was obtained with phosphorylase 6 , indicating that the dephosphorylated Nterminus could not prevent the phosphorylated peptide from binding, again em-


38 1

phasizing the difference between the a and b forms with respect to the conformation of the N-terminus. Finally, Janski and Graves (106)showed that nucleotide improved markedly the binding of a synthetic phosphorylated peptide (residues 1-18) to phosphorylase b. Surprisingly, in view of the crystal structure, studies with various lengths of synthetic peptide suggested that residues 1-4 improve the binding, whereas residues 5-6 (Ser-Asp) are very important. This is consistent with unpublished data from crystallographic studies which suggest that the Nterminus binds more extensively as well as more strongly in the R-state than in the T-state.* Although the actual amount of extra energy of stabilization of the phosphorylase a dimer (compared to the b dimer) that has been conferred by the two new intersubunit salt bridges is the subject of some debate (76),the literature provides ample qualitative evidence for the existence of this stabilization. Thus phosphorylase a is more stable to thermal denaturation (107).Evidence has been provided that phosphorylase b dimers dissociate to a small extent and, at low concentrations, are completely dissociated in 2 M NaCl, whereas phosphorylase a is merely dissociated from tetrameric to dimeric form in that concentration of salt (108). The b forms of the heart-muscle phosphorylase isozymes hybridize readily whereas the a forms do not (47), and this result is pertinent to our present discussion because heart isozyme 111 appears to be identical to the single isozyme of skeletal muscle. Interestingly, enough, AMP or glucose-6-P were, like phosphorylation of serine-14, able to block hybridization. In another relevant study, Shaltiel et al. showed that 0.4M imidazole citrate causes the dissociation of phosphorylase b into monomers (109), from which the coenzyme is readily resolved with cysteine, whereas this deforming salt does not dissociate phosphorylase a (110).Again, either AMP or phosphorylation of serine-14 prevents resolution of the pyridoxal phosphate in the imidazole citrate buffer (104). The stabilization of subunit interactions by AMP or glucose-6-P may be explained by their interaction with residues in both subunits (58, 71). One more example of the subunit association in phosphorylase may be given to indicate that there remain unresolved ambiguities. Hedrick ef al. (111) showed that apophosphorylaseb exists as a tetramer at 0", a dimer at 23",and amonomer at 35".A later study showed much the same behavior for apophosphorylasea (112), so that we are unable to distinguish between the phosphorylated and unphosphorylated forms of apophosphorylase with respect to intersubunit stability. This appears to be the only exception to evidence suggesting the greater stability of the phosphorylase a dimer. While removal of the coenzyme destabilizes the intersubunit associations, it has little effect on some other parameters since AMP can still bind (111), glycogen binds as well as to the holoenzyme (113), and phos*Personal communications from Dr. R. J. Fletterick, reported also in his address before "The Robert A. Welch Foundation Conferences on Chemical Research," XVII, Houston, Texas (1983).

382

NEIL B. MADSEN

phorylase kinase exhibits the same rate of reaction on the apo form of phosphorylase b as on the holo form (1 11). Thus the loss of the coenzyme does not appear to alter significantly the structure of the N-terminus or of the glycogen storage site. However, the accessibility of the serine- 14 phosphate may be greater in apophosphorylase a as judged by the increased activity of the protein phosphatase with this substrate (96). As previously mentioned, AMP is able to inhibit phosphatase action on the apophosphorylase a whereas there is no effect of glucose or caffeine, again emphasizing the importanceof intersubunitcontacts for the transmital of conformational changes between the active site region and the regulatory region containing the sites for serine phosphate and AMP. The two serine-14 phosphates of the phosphorylase a dimer are separated by approximately 40 8, around the curved surface of the protein, and it was suggested that the huge multimeric protein kinase might be able to phosphorylate both serine- 14s simultaneously (62). Similarly, the “holophosphatases,” presumably containing more than one catalytic subunit, may also be able to carry out a simultaneous dephosphorylation. Nevertheless, convincing evidence was obtained that partially phosphorylated intermediates occur during the interconversion of phosphorylase a and b (114, 115). Thus, during the phosphatase reaction, 50% of the phosphate could be released from serine- 14 with no loss of activity as measured at high concentrations of glucose- 1-P whereas, when measured at the usual low concentrations of substrate, or in the presence of glucose-6-P, the loss of activity coincided with dephosphorylation. The reverse phenomenon was observed during the conversion of phosphorylase b to a by the kinase. The results were interpreted as indicative of the formation of phosphorylase alb hybrids having properties intermediate with respect to the parent forms, so that the affinity for glucose-1-P was less than that for a but greater than for b. Similarly, the inhibition by glucose-6-P is reminiscent of the b form, not the a. Two serine phosphates per tetramer stabilized the tetrameric state in the presence of glucose-1-P (115) while the hybrid dimer showed a reversal of caffeine inhibition by AMP which the phosphorylase b dimer does not (116). These results are reasonably interpreted, in light of the structures of phosphorylase a and b, as suggesting that the phosphorylase alb dimer has one phosphorylated N-terminus bound across the subunit interface, providing part of the increased stabilization and site-site interaction observed upon full conversion to the phosphorylated form. Randomization (reshuffling) of the two types of subunits in the oligomers, as well as formation of hybrids from the homogeneous forms, has also been observed and analyzed (115). The suggestion that hybrid forms may occur in vivo during the interconversions of the a and b forms and thus lead to increased sensitivity to glucose-6-Pmediated regulation of glycogenolysis has received some support by the finding of these hybrids during the “flash activation” of phosphorylase bound to glycogen particles (117). In addition, the conversion of phosphorylase b to a in


383

rabbit and frog muscle in response to hormonal and electrical stimulation was investigated by using activity measurements sensitive to the hybrid species (118). The assay for the hybrid species was based on these workers’ previous finding that AMP increases its activity in the presence or absence of caffeine whereas phosphorylase b activity in the presence of AMP is completely inhibited by caffeine (116). They also showed that AMP could reverse the ATP inhibition of the hybrid under conditions where this is not true for phosphorylase b. Radda and his colleagues at Oxford employed electron spin resonance techniques with spin-labelled phosphorylase b to demonstrate the transient formation of the alb hybrid as an intermediate in the in vitro conversion (8). However. they were unable to detect the intermediate during conversion of phosphorylase b to a in the isolated glycogen particle. It is apparent that the formation of phosphorylase alb hybrids under normal physiological conditions is a distinct possibility, and this species may exhibit different control characteristics from either of the more stable forms, as suggested by the two groups most active in this area (114, 118). One feels that further investigation of these phenomena is warranted, both to confirm the formation of the hybrids under physiological conditions by using more direct chemical methods, and to explore in more detail their significance with respect to regulation. This is an intriguing subject with considerable interest both for the regulation of glycogen phosphorylase and as a model for other more complicated protein phosphorylation systems.

B, OTHERMEASURABLE STRUCTURAL EFFECTS One of the most notable results of the phosphorylation of serine-14 is the association of dimers to form tetramers in the case of phosphorylases from most skeletal muscles, as discussed in Section III,A. This association must be related to the N-terminal region but no information is available as to the actual residues involved, and there is sufficient evidence for the long-range transmission of conformational changes in this molecule to caution us against assuming that the N-terminus forms the interface. Phosphorylase b will form tetramers in the presence of AMP, especially with Mg, while phosphorylase b’, lacking the first 16 residues, will not (105). Huang and Graves (119) determined the dissociation constant for the tetramer to dimer dissociation and showed that it increased markedly with temperature, yielding standard enthalpy changes of 60 kcal/mol and an entropy change of 170 units. Many of the differences observed in the reactivities of functional groups of the two forms of phosphorylase may well be due to the difference in oligomeric state. The rates for both the inhibition and the subsequent dissociation of phosphorylase b as a consequence of mercurials reacting with sulfhydryl groups is

384

NEIL B. MADSEN

slower than those observed with phosphorylase a (44, 119, 120). The same phenomenon was observed when iodoacetamide was employed (110, 121). In this case specific rate constants for Cys-108 and Cys-142 were measured and those in phosphorylase a reacted faster than those in phosphorylase 6. Access to Cys-142 is via the internal cavity between the two monomers, and requires some flexing or “breathing.” Because the cavity is closed to the external solvent in the static T-state (76),the cysteine in the less tightly associated b dimer would be expected to have a greater reactivity than that in the more rigid a tetramer. Clearly, we do not yet understand all the factors involved in these structurefunction relationships.

VI. Functional Results of the Phosphorylation of Serine-14 A. CHANGES IN THE ALLOSTERICCONSTANT, L, AND IN AMP BINDING Although it is generally considered to be an over simplification to apply the concerted model for allosteric transitions to phosphorylase b, nevertheless it is useful to use the concept of a “final” inactive T-state conformation and an active R-state conformation. The allosteric constant, L, which is the equilibrium constant for the molar ratio of T to R, has been estimated to be at least 3300 (122, 123). The same constant L for phosphorylase a, where the two-state model is more applicable, has been estimated at between 3 and 13 (8, 124). For discussion purposes we may assign nominal values of 3000 and 10, indicating that the energy required for the T +-R transition has been reduced by at least 3.5 kcal. This reduction in energy must come from the interaction of the serine-14 phosphate with the main body of the dimer, as discussed in Section V,A. The strength of the newly formed salt bridges, plus other possible interactions, remains a matter for lively discussion (76),with values for salt bridges estimated at anywhere from 1 kcal/mol in hemoglobin (125) to 3 for an internal salt bridge in chymotrypsin (126).The maximal value for a salt bridge formed from a proteinbound serine phosphate that is accessible to solvent has been estimated at 5 kcal (100). For purposes of discussion, a nominal value of 4 kcal has been assigned to represent the net energy of stabilization afforded by the interaction of the phosphorylated N-terminus across the subunit interface of the phosphorylase dimer. In Fig. 3 an attempt has been made to analyze the thermodynamic consequences of the phosphorylation of serine-14. As previously pointed out the effect on the allosteric constant L, reduces the energy of this transition from +4.9 to + 1.4 kcal. This reduction coincides almost exactly with the increased energy of binding of AMP. Thus the latter binds 200 times more tightly to phosphorylase a than to b ( K , = 2 pA4 versus 400 pA4), increasing the binding energy by 3.2 kcal

385


P

FIG. 3. Thermodynamic analysis of the effect of phosphorylation on allosteric transitions in the phosphorylase system. Phosphorylase b dimers are shown on the left, phosphorylation of serine-14 and the subsequent binding of the N-terminus across the subunit interface is depicted on top horizontal line whereas phosphorylase a dimers are shown on the right. Circles represent the T conformation and squares the R conformation, but dimers liganded with AMP will have a different conformation than unliganded dimers. Numbers are estimates for the free energy changes (in kcal) for each equilibrium.

(-4.7 to -7.9 kcal), a value close to but somewhat smaller than the change in the energy of the allosteric transition. It is obvious that, in phosphorylase b, a large fraction of the binding energy of the activating nucleotide is used to cause a conformational change leading from the T- to the R-state. In the case of phosphorylase a, although not symbolized on the diagram, the conformation of the AMP-liganded enzyme is very close to the R-state, and, conversely, substrate alone is able to stabilize a conformation similar to the R-state, as shown by identical maximal velocities in the presence and absence of AMP. On the contrary, it was first shown by Radda and colleagues, reviewed in (8), that glucose-1-P induced a major conformational change in AMP-liganded phosphorylase b. Part of the energy for the T + R transition in phosphorylase b must therefore be provided by the binding of the substrate and the crude estimate for the second part of this transition, shown in the figure as +2.3 kcal, is based, perhaps naively, on the limiting K,,, for glucose- 1-P for phosphorylase b being 10 times that for phosphorylase a (22). Therefore, 1.3 kcal was added to the 1.0 estimated as the maximal energy involved in the similar transition of AMPliganded phosphorylase a to the fully activated R-state saturated with substrates. The diagram shown in Fig. 3 is an oversimplification and the values given for the various free energies should not be taken as anything but estimates. The purpose is to illustrate how, as Sprang and Fletterick suggest, “we can imagine the N-terminus to behave as an intramolecular allosteric effector of the R-state

386

NEIL B . MADSEN

organizing the subunit surface. This ‘effector’ must be phosphorylated to bind at the dimer surface” (76). The resultant symmetrical alterations in compensatory free energy changes are more than just coincidential. The considerable increase in subunit interactions within the dimer that is brought about by the binding of the phosphorylated N-terminus across the subunit interface was discussed previously. Not surprisingly, this has made the existence of intermediate conformations involving different conformations of the two subunits much more difficult in phosphorylase a , so that symmetry must be conserved, as required by the concerted transition model of Monod et al. (127). Thus Griffiths et al. (128) have shown that the binding of AMP to phosphorylase a follows the symmetry model rather than the simple sequential model. On the other hand, members of the same group, as reviewed by Busby and Radda (8), have shown that the binding of most ligands to phosphorylase b follows the sequential model of Koshland et al. (129). As reviewed by Madsen et al. (7) other laboratories have produced data that tend to agree, for the most part, with these conclusions, although there are some rather untidy discrepancies. As Koshland pointed out, the concerted transition or symmetry model is a limiting case of the general sequential model and may be brought about when the intersubunit interactions are too strong to permit stable intermediate forms (130). The phosphorylase system represents, at least to a first approximation, an example of how a simple chemical modification provides an enhanced subunit interaction sufficient to transform sequential allosteric transitions to the concerted mode. CONTROL B. ESCAPEFROM ALLOSTERIC This heading is meant to dramatize the fundamental biological significance of the conversion of phosphorylase b to a under physiological conditions, and not to imply that the a form does not exhibit allostery. Figure 4 demonstrates that phosphorylase a has lost its requirement for AMP under conditions where phosphorylase b exhibits a virtually obligate requirement for activity (Shechosky and Madsen, unpublished research). It may be seen, too, that ATP causes severe inhibition of phosphorylase b, while accentuating the homotropic interaction of the AMP-binding sites. On the other hand, ATP inhibits only that extra activity exhibited by phosphorylase a in the presence of AMP. The data illustrate also the much greater binding affinity of phosphorylase a for AMP. Thus the phosphorylation of serine-14 eliminates both the need for an activating nucleotide and the inhibition by metabolites that bind at the activator site, namely ATP, ADP, and glucose-6-P. Lowry et al. pointed out that since the main effect of AMP on phosphorylase a was to decrease the K , for the substrates, glucose-1-P, and inorganic phosphate, the activation would be important at the low Pi concentrations found in muscle (131). However, the apparent K,,,for Pi in the absence of AMP is 3 mM so that

387


1.0

r

0.61 Y

f

P

I

50

100

p M AMP

150

boo

3

FIG. 4. The effects of AMP on the activity of phosphorylases a and b, as modulated by ATP. Activity was measured at 30” with 16 mM glucose-I-P and 1 % glycogen and is represented by the ratio to maximal activity. When present, ATP was 9 mM; a indicates results with phosphorylase a, b represents phosphorylase b, and n is the Hill coefficient (from S. Shechosky and N. B . Madsen, unpublished experiments).

r,

the enzyme would have partial activity with this substrate in the millimolar range (22). Busby and Radda addressed the problem of control of phosphorylase activity in activated muscle and conclude that if it were all converted to the a form, half would be liganded with AMP, the remainder with ATP and ADP (8).While maximal activity would not be expected under these conditions, the enormous concentration of phosphorylase in muscle would be more than adequate to account for the observed activities, the latter being directly related to phosphate concentration. On the other hand, phosphorylase b in the resting muscle would be expected to be almost totally liganded with glucose-6-P, ATP, and ADP, with almost no bound AMP (8),thus accounting for the very low rate of glycogenolysis under these conditions (132). In 1962, Danforth et al. (133)developed a method to “freeze” instantly the interconverting system for phosphorylase in muscle, and to extract the enzymes and measure their activities under conditions which precluded any further changes. They were thus able to establish that the phosphorylase a activity of resting frog sartorius muscle is less than 5% of the total potential activity, thereby resolving the previous unsatisfactory relationship between the rate of glycogenolysis and the phosphorylase a content (132).They were able to establish the kinetics of the activation of phosphorylase a in response to either electrical stimulation or adrenalin, the effect of the latter being much slower, as well as the kinetics for the decay of the phosphorylase a activity after stimulation. These studies have been summarized by Helmreich and Cori (134).

388

NEIL B. MADSEN

A major contribution to the studies of the interconversion of phosphorylase b and a in muscle was made possible by the discovery of a strain of mice that appear to lack phosphorylase b kinase (135). Danforth and Lyon (136) demonstrated that while the proportion of phosphorylase in the a form in resting muscles of these I-strain mice was the same as that in the muscles of normal C57 mice, approximately 5%, electrical stimulation of the latter resulted in a rapid transformation of phosphorylase to 70% a (with a half-time of one second), whereas no effect was seen in the muscles from the I-strain mice. Furthermore, there was a pronounced lag in the onset of glycogenolysis in the muscles from the I-strain mice compared to the normal situation, as measured by the production of glucose-6-P or lactate. The rate of formation of these two compounds was half that by the normal muscles, and the total produced was also half. Therefore the importance of the conversion of phosphorylase b to a was well illustrated but the question of a mechanism that would allow the stimulation of phosphorylase b during muscle contraction was also raised by these studies. The most obvious explanation, an increase in the concentration of AMP sufficient to activate phosphorylase b, was discounted by the studies of Griffiths and his colleagues, who showed significant changes in the AMP levels of muscles from normal or phosphorylase kinase-deficient mice upon 15-30 min of forced exercise by swimming (137).Furthermore, while the total AMP concentration was of the order of 0.1 mM, they suggested on the basis of the adenylate kinase equilibrium that the free AMP concentration might not exceed 0.006 mM, too low to activate phosphorylase b. On the other hand, the IMP concentrations rose from 0.73 mM in the muscles of the resting kinase-deficient mice to 1.65 mM after exercise. Since the K , of IMP for phosphorylase b is less than 1 mM (138), this could provide sufficient activation to account for the increased glycogenolysis. One problem with scenario revolves around the earlier finding of Black and Wang (138) that IMP does not improve the K,,, for the substrate, glucose-1-P, which remains at at least 32 mM at all IMP concentrations tested. We have confirmed (unpublished experiments) that in the direction of phosphorolysis at 1 mM IMP, the concentration of Pi required to reach half-maximal velocity is approximately 40 mM, the Hill coefficient is 1.7 and the maximal velocity (at 10 pmol/min/mg) is only one-third that observed with AMP. Furthermore, at 5 mM Pi, the specific activity was only 0.27 pmol/min/mg. C. CHARACTERISTICS REMAINING UNCHANGED

It may be unnecessary to belabor the obvious fact that most of the basic enzymic characteristics of phosphorylase are not altered by phosphorylation of serine-14, but since this is the prototype interconvertible enzyme system, principles established with it may act as guides for evaluating other systems. There is no reason to believe that the nature of the catalytic mechanism has been altered in


389

any fundamental aspect. While the maximal activity of phosphorylase a is slightly more than that of the b form, this probably does not reflect any change in the catalytic mechanism. The energy of activation of the two forms, measured in the presence of AMP, is identical within experimental error (7). The two forms bind to glycogen with much the same dissociation constant, indicating a limited effect of the N-terminus on the glycogen storage site and thereby ensuring that phosphorylase remains bound to the glycogen particle regardless of its state of activity. Another characteristic which varies only in detail between the two forms is the synergistic inhibition by glucose and caffeine. The kinetically derived dissociation constants for glucose with the a and b forms are approximately 3 and 1 mM, respectively, while those for caffeine are 0.2 and 0.08 mM, respectively (64, 66). The interaction constants are quite similar, being calculated at 0.3 and 0.2 for the a and b forms. Because of the uncertainties in these calculations, there is no significant difference in the extent to which one of these ligands improves the binding of the other. Although the ligands bind somewhat more tightly to phosphorylase b than to a, the conversion has not eliminated the inhibitions or modified the allosteric interactions. Any metabolic controls exerted at or near the catalytic site may possibly be affected only slightly by covalent interconversion. It will be interesting to see if this is a general principle applicable to other metabolically interconvertible enzymes.

VII. Concluding Remarks The control of skeletal muscle glycogen phosphorylase by the reversible phosphorylation of its serine-14 provides us with a model system in which we should be able to discern principles of structure-function relationships applicable to other metabolically interconvertible enzymes. We have seen that after phosphorylation the N-terminal 18 residues bind across the subunit interface and strengthen the subunit interactions. It is suggested that the extra energy afforded by this interaction reduces the energy required for allosteric transitions and thereby allows phosphorylase a to escape from the allosteric controls to which phosphorylase b is subject. Many other interconvertible enzyme systems exhibit a similar release from allosteric restraints in their active forms, even though activation may involve dephosphorylation rather than phosphorylation of critical residues, and we may look forward, in due course, to the delineation of structural changes similar in principle to those discovered for the phosphorylase system. Increasing complexity will be observed in most cases, however, because, while phosphorylase is complicated enough, it is a “clean” and simple enzyme compared to some other systems. Just to take examples from the area of glycogen metabolism, glycogen synthase exhibits the phosphorylation of seven serines

390

NEIL B. MADSEN

arranged in three groups with various phosphorylation patterns exhibited as the result of action by three separate kinases. Phosphorylase b kinase is composed of four different types of subunits, two of which undergo phosphorylation. While progress in the structure-function relationships in the phosphorylase system may seem impressive, we are at a rather superficial level in our understanding and considerable work is required to clarify our working hypothesis. Comparison of the refined high-resolution structures of the a and b forms of phosphorylase is planned and should be very useful. Since both structures are of the inactive T conformation, we need the structure of at least one R-form so we can understand how serine-14 phosphorylation facilitates the T + R transition. We have only the slightest knowledge of how the enzymes catalyzing the interconversion interact with their phosphorylase substrates. As pointed out in Section VI,B, research on mice that lack phosphorylase b kinase has illuminated the physiological benefits of the conversion from the b to a form, but the mechanism by which these mice can still carry out glycogenolysis remains a matter for further investigation. The reader will note that I have emphasized the skeletal muscle phosphorylase system while saying little about the equally important system in liver, let alone those in other tissues. Aside from feeling more comfortable with the muscle system because of its wealth of structure-function information, I consider that the state of research on the liver system is in great flux, making a definitive treatment difficult. One should encourage a concerted effort to determine structures for the phosphorylases of liver because this would provide a firm base for defining the physiological role and mechanism for their control by phosphorylation. ACKNOWLEDGMENTS I am grateful to Dr. R. J . Fletterick for helpful discussions and for providing Fig. I . I wish to thank Mrs. P. McDonald for her skillful typesetting through several revisions. Research reported from this laboratory was supported by Grant MT-1414 from the Medical Research Council of Canada.

REFERENCES 1. 2. 3. 4.

Green, A . A , , Cori, G. T . , and Cori, C. F. (1942). JBC 142, 447-448. Fischer, E. H., and Krebs, E. J . (1955). JBC 216, 121-132. Krebs, E. G., and Fischer, E. H. (1956). BBA 20, 150-157. Fischer, E. H., Graves, D. J . , Crittenden, E. R. S . , and Krebs. E. H. (1959). JBC 234, 16981704. 5. Fischer, E. H . , Graves, D. H., and Krebs, E. G. (1957). FP 16, 180. 6. Graves, D. J . , and Wang, J . H. (1972). “The Enzymes,” 3rd ed., Vol. 7, pp. 435-482.


39 1

7. Madsen, N. B., Avramovic-Zikic, 0.. Lue, P. F., and Honkel. K. 0. (1976). Mol. Cell Biochem. 11, 35-50. 8. Busby, S. J. W., and Radda, G. K. (1976). Curr. Top. Cell. Regul. 10, 89-160. 9. Helmreich, E. J., and Klein, H. W. (1980). Angew. Chem. 19, 441-455. 10. Fletterick, R. J., and Madsen, N. B. (1980). Annu. Rev. Biochem. 49, 31-61. 11. Dombradi, V. (1981). Inr. J . Biochem. 13, 125-139. 12. Madsen, N. B., and Withers, S. G. (1986). In “Coenzymes and Cofactors. Pyridoxal Phosphate and Derivatives” (D. Dolphin, R. Poulson, and 0. Avramovic, eds.), Vol. IB, pp. 355385. Wiley, New York. 13. Cori, G. T . , and Con, C. F. (1940). JBC 135, 733-756. 14. Lamer, J., Villa-Palasi, C., and Rechman, D. J. (1960). ABB 86, 56-60. 15. Taylor, C., Cox, A. J.. Kevrohan, J. C., and Cohen, P. (1975). EJB 51, 105-115. 16. Wanson, J.-C., and Drochmans, P. (1968). JBC 38, 130-150. 17. Porter, K. R., and Bruni, C. (1959). Cancer Res. 19, 997-1009. 18. Meyer, F., Heilmeyer, L. M. G., Haschke, R. H., and Fischer, E. H. (1970). JBC 245,66426647. 19. Caudwell, B., and Cohen, P. (1978). EJB 86, 511-518. 20. Maddaiah, V. T., and Madsen, N. B. (1966). JBC 241, 3873-3881. 21. Engers, H. D., Bridger, W. A., and Madsen, N. B. (1969). JBC 244, 5936-5942. 22. Engers, H. D., Shechosky, S., and Madsen, N. B. (1970). Can. J . Biochem. 48, 746-754. 23. Engers, H. D., Bridger, W. A., and Madsen, N. B. (1970). Can. J. Biochem. 48, 755-758. 24. Gold, A. M., Johnson, R. M., and Tseng, J. K. (1970). JBC 245, 2564-2572. 25. Engers, H. D., Bridger, W. A , , and Madsen, N. B. (1970). Biochemistry 9, 3281-3284. 26. Madsen, N. B., and Withers, S. G. (1984). I n “Chemical and Biological Aspects of Vitamin B6 Catalysis” (A. E. Evangelopolous, ed.), Part A, pp. 117-126. Alan R. Liss, Inc., New York. 27. Klein, H. W., Im, M. J., and Helmreich, E. J. M. (1984). In “Chemical and Biological Aspects of Vitamin B6 Catalysis (A. E. Evangelopolous, ed.), pp. 147-160. 28. Sygusch, J., Madsen, N. B., Kasvinsky, P. J., and Fletterick, R. J. (1977). PNAS 74, 47574761. 29. Weber, I. T., Johnson, L. N., Wilson, K. S . , Yeates, D. G. R., and Wild, D. I . (1978). Nature (London) 274, 433-436. 30. Parrish, T., Uhing, R. J., and Graves, D. J. (1977). Biochemistry 16, 4824-4831. 31. Withers, S . G., Madsen, N. B., Sykes, B. D., Takagi, M., Shimomura, S., and Fukui, T. (1981). JBC 256, 10759-10762. 32. Klein, H. W., Palm, D., and Helmreich, E. J. M. (1982). Biochemistry 21, 6675-6684. 33. Takagi, M., Fukui, T., and Shimomura, S. (1982). PNAS 79, 3716-3719. 34. Withers, S. G., Madsen, N. B., Sprang, S. R., and Fletterick, R. J. (1982). Biochemistry 21, 5372-5382. 35. Avramovic-Zikic, O., Breidenbach, W. C., and Madsen, N. B. (1974). Can. J. Biochem. 52, 146- 148. 36. Shimomura, S., Nakano, K., and Fukui, T. (1978). BBRC 82, 462-468. 37. Dreyfus, M., Vandenbunder, B., and Buc, H. (1980). Biochemisrry 19, 3634-3642. 38. Kasvinsky, P. J., and Meyer, W. L. (1977). ABB 181, 616-631. 39. Klein, H. W., Schiltz, E., and Helmreich, E. J. M. (1981). In “Protein Phosphorylation” (0. M. Rosen and E. G. Krebs, ed.), pp. 305-320. Cold Spring Harbor Lab., Cold Spring Harbor, New York. 40. Cori, C. F., and Cori, G. T. (1936). Proc. SOC. Exp. B i d . Med. 34, 702-705. 41. Con, C. F., Cori, G. T., and Green, A. A. (1943). JBC 151, 39-55.

392

NEIL B. MADSEN

Parmeggiani, A., and Morgan, H. E. (1962). BBRC 9, 252-256. Keller, P. J . , and Cori, G. T. (1953). BBA 12, 235-238. Madsen, N. B., and Cori, C. F. (1956). JBC 223, 1055-1065. Appleman, M. M., Krebs, E. G., and Fischer, E. H. (1966). Biochemistry 5, 2101-2107. Assaf, S. A., and Graves, D. J . (1969). JBC 244, 5544-5555. Davis, C. H . , Schlisefield, L. H., Wolf, D. P., Leavitt, C. A,, and Krebs, E. G. (1967). JBC 242, 4824-4833. 48. Kent, A. B., Krebs, E. G., and Fischer, E. H. (1958). JBC 232, 549-558. 49. Huang, C. Y., and Graves, D. J. (1970). Biochemistry 9, 660-671. 50. Titani, K., Koide, A., Ericsson, L. H., Kumar, S., Wade, R., Walsh, K. A., Neurath, H., and Fisher, E. (1977). PNAS 74, 4762-4766. 51. Palm, D., Goerl, R., Burger, K. J., Buhner, M., and Schwartz, M. (1984).In “Chemical and Biological Aspects of Vitamin 8 6 Catalysis” (A. E. Evangelopoulos, ed.), pp. 209-221. Alan R. Liss, Inc., New York. 52. Nakano, K., Kikumoto, Y., and Fukui, T. (1984). In “Chemical and Biological Aspects of Vitamin 86 Catalysis” (A. E. Evangelopoulos, ed.), pp. 171-180. Alan R. Liss, Inc., New York. 53. Lerch, K., and Fischer, E. H. (1975). Biochemistry 14, 2009-2014. 54. Cohen, P., Saari, J . C., and Fischer, E. H. (1971). Biochemistry 10, 5233-5241. 5 5 . Jenkins, L. N., Stuart, D. I., Stura, E. A., Wilson, K. S., and Zanotti, G. (1981). Philos. Trans. R . SOC. London, Ser. B 293, 23-41. 56. Lorek, A,, Wilson, K. S., Stura, E. A., Jenkins, J . A,, Zanotti, G., and Johnson, L. N. (1980). JMB 140, 565-580. 57. Sansom, M. S. P., Stura, A., Babu, Y. S., McLaughlin, P., and Johnson, L. N. (1984). In “Chemical and Biological Aspects of Vitamin B6 Catalysis” (A. E. Evangelopoulos, ed.), pp. 127-146. Alan R. Liss, Inc., New York. 58. Lorek, A., Wilson, K. S., Sansom, M. S. P., Stuart, D. I., Stura, E. A,, Jenkins, J . A., Hajdu, J . , and Johnson, L. N. (1984). BJ 218,45-60. 59. Sprang, S. R . , and Fletterick, R. J. (1979). JMB 131, 523-551. 60. Sprang, S. R . , Goldsmith, E. J., Fletterick, R. J., Withers, S. G., and Madsen, N. B. (1982). Biochemistry 21, 5364-5371. 61. Madsen, N. B., Kasvinsky, R. J., and Fletterick, P. J . (1978). JBC 253, 9097-9101. 62. Fletterick, R. J . , Sprang, S., and Madsen, N. B. (1979). Can. J. Biochem. 57, 789-797. 63. Kasvinsky, P. J . , Madsen, N. B., Sygusch, J., and Fletterick, R. J . (1978). JBC 253, 33433351. 64. Kasvinsky, P. J., Shechosky, S., and Fletterick, R. J . (1978). JBC 253, 9102-9106. 65. Withers, S. G., Sykes, B. D., Madsen, N. B., and Kasvinsky, P. J . (1979). Biochemistry 24, 5342. 66. Madsen, N. B., Shechosky. S., and Fletterick, R. J. (1983). Biochemistry 22, 4460-4465. 67. Sygusch, J . , Madsen, N. B., and Fletterick, R. J . (1977). PNAS 74, 4757-4761. 68. Johnson, L. N., Jenkins, J. A., Wilson, K. S., Stura, E. A,, and Zanotti, G. (1980).JMB 140, 565-580. 69. Withers, S. G., Madsen, N. B., Sprang, S. R., and Fletterick, R. J. (1982). Biochemistry 21, 5372-5382. 70. Sprang, S. R . , Fletterick, R. J., Stem, M., Yang, G., Madsen, N. B., and Sturtevant, I. M. (1982). Biochemistry 21, 2036-2048. 71. Stura, E. A., Zanotti, G., Babu, Y. S., Sansom, M. S. P., Stuart, D. I., Wilson, K. S., and Johnson, L. N. (1983). JMB 170, 529-565. 72. Lee, Y. M . , and Benisek, W. F. (1976). JBC 251, 1553-1560. 73. Lee, Y. M., and Benisek, W. F. (1978). JBC 253, 5460-5463.

42. 43. 44. 45. 46. 47.


393

74. Johnson, L. N., Stura, E. A,, Wilson, K. S., Sansom, M. S. P., and Weber, I. T. (1979). JME 134, 639-653. 75. Weber, I. T., Johnson, L. N., Wilson, K. S., Yeates, D. G. R., Wild, D. L., and Jenkins, J. A. (1978). Nature (London) 274, 433-437. 76. Sprang, S., and Fletterick, R. J. (1980). Eiophys. J. 32, 175-192. 77. Krebs, E. G., Love, D. S., Bratvold, G. E., Traysey, K.A,, Meyer, W. L., and Fischer, E. H. (1964). Biochemistry 3, 1022-1033. 78. Tabatabai, L. B., and Graves, D. J. (1978). JEC 253, 2196-2202. 79. Tu, J. I., and Graves, D. J. (1973). EERC 53, 59-65. 80. Gusev, N. B., and Hajdu, J. (1979). EERC 90, 70-77. 81. Ingebritsen, T. S., Stewart, A. A,, and Cohen, P. (1983). EJE 132, 297-307. 82. Ingebritsen, T. S., Foulkes, J. G., and Cohen, P. (1983). EJE 132, 263-274. 83. Martensen, T. M., Brotherton, J. E., and Graves, D. J. (1973). JEC 248, 8323-8328. 84. Martensen, T. M., Brotherton, J. E., and Graves, D. J. (1973). JEC 248, 8329-8336. 85. Detwiler, T. C., Gratecos, D., and Fischer, E. H. (1977). Biochemistry 16, 4818-4823. 86. Nolan, C., Nova, W. B., Krebs, E. G., and Fischer, E. H. (1964). Biochemistry 3,542-551. 87. Sutherland, E. W. (1951). In “Phosphorus Metabolism” (W. S. McElroy and B. Glass, eds.), pp. 53-61. Johns Hopkins Press, Baltimore, Maryland. 88. Withers, S. G., Madsen, N. B., and Sykes, B. D. (1981). Biochemistry 20, 1748-1756. 89. Dombradi, V., Toth, B., Bot, G., Hajdu, J., and Friedrich, P. (1982). fnt. J . Eiochem. 14, 277-284. 90. Graves, D. J . , Mann, S. A. S., Philip, G., and Oliveira, R. J. ( I 968). JEC 243, 6090-6098. 91. Stalmans, W., Laloux, M., and Hers, H. G. (1974). EJE 49, 415-427. 92. Hers, H. G. (1976). Annu. Rev. Eiochem. 45, 167-189. 93. Stalmans, W. (1976). Curr. Top. Cell. Regul. 11, 51-97. 94. Kasvinsky, P. J., Fletterick, R. J . , and Madsen, N. B. (1981). Can. J. Biochem. 59,387-395. 95. Monanu, M. 0.. and Madsen, N. B. (1985). Can. J. Eiochem. CellEiol. 63, 115-121. 96. Yan, S. C. B., Uhing, R. J., Parrish, R. F., Metzler, D. E., and Graves, D. I. (1979). JEC 254, 8263-8269. 97. Bot, G., Kovacs, E., and Gergely, P. (1977). Acia Eiochim. Eiophys. Acad. Sci. Hung. 12, 335-341. 98. Raibaud, O., and Goldberg, M. E. (1973). Eiochemisrry 12, 5154-5161. 99. Dombrhdi, B., Tbth, B., Gergely, P., and Bot, G. (1983). Int. J. Biochem. 15, 1329-1336. 100. Vogel, H. J., and Bridger, W. A. (1983). Can. J . Eiochem. CeNEiol. 61, 363-369. 101. Sealock, R. W., and Graves, D. J. (1967). Biochemistry 6, 201-207. 102. Janski, A. M., and Graves, D. J. (1979). JEC 254, 1644-1652. 103. Cori, G. T., and Con, C. F. (1945). JEC 158, 321-332. 104. Keller, P. J. (1955). JEC 214, 135-141. 105. Carty, T. J . , Tu, J.-I., and Graves, D. J. (1975). JEC 250, 4980-4985. 106. Janski, A. M., and Graves, D. J. (1979). JEC 254, 4033-4039. 107. Graves, D. J., Sealock, R. W., and Wang, J. H. (1965). Eiochernistry4, 290-296. 108. Cohen, P., Duewer, T . , and Fischer, E. H. (1971). Biochemistry 10, 2683-2694. 109. Shaltiel, S., Hedrick, J . L., and Fischer, E. H. (1966). Biochemistry 5, 2108-21 16. 110. Avrarnovic, 0.. Smillie, L. B., and Madsen, N. B. (1970). JEC 245, 1558-1565. 1 1 I . Hedrick, J. L.. Shaltiel, L., and Fischer, E. H. (1966). Biochemistry 5, 21 17. 112. Shaltiel, S., Hedrick, J. L., Pocker, A,, and Fischer, E. H. (1969). Biochemistry 8, 51895196. 113. Kastenschmidt, L. L., Kastenschmidt, J., and Helmreich, E. (1968). Biochemistry 7 , 35903608. 114. Hurd, S. S., Teller, D., and Fischer, E. H. (1966). EBRC 24, 79-84.

394

NEIL B. MADSEN

115. Fischer, E. H., Hurd, S . S . , Koh, P., Seery, V. L., and Teller, D. C. (1968). In “Control of Glycogen Metabolism” (W. J. Whelan, ed.), pp. 19-33. Academic Press, New York. 116. Bot, Y.,Kovics, E. F., and Gergely, P. (1974). BBA 370, 78-84. 117. Heilmeyer, L. M. G.,Meyer, F., Haschke, R. H., and Fischer, E. H. (1970). JBC 245,66496656. 118. Gergely, P., Bot, G.,and Kovhcs, E. F. (1974). BBA 370, 78-84. 119. Huang, C. Y., and Graves, D. J. (1970). Biochemistry 9, 660-671. 120. Madsen, N. B. (1956). JBC 223, 1067-1074. 121. Battell, M. L., Zarkadas, C. G.,Smillie, L. B., and Madsen, N. B. (1968). JBC 243,62026209. 122. Madsen, N. B., and Shechosky, S . (1967). JBC 242, 3301-3307. 123. Kastenschmidt, L. L., Kastenschmidt, J., and Helmreich, E. (1968). Biochemisrry 7 , 45434550. 124. Helmreich, E., Michaelides, M. C., and Con, C. F. (1967). Biochemistry 6 , 3695-3710. 125. Perutz, M. F. (1978). Science 201, 1187-1 191. 126. Fersht, A. R. (1971). Cold Spring Harbor Symp. Quant. Biol. 36, 71-73. 127. Monod, J . , Wyman, J . , and Changeux, J.-P. (1965). JMB 12, 88-1 18. 128. Griffiths, J. R., Price, N. C., and Radda, G. K. (1974). BBA 358, 275-280. 129. Koshland, D. E., Nemethy, G.,and Filmer, D. (1966). Biochemistry 5, 365-385. 130. Koshland, D. E. (1969). Curr. Top. Cell. Regul. 1, 1-27. 131. Lowry, 0. H., Schulz, D. W., and Passonneau, J. V. (1964). JBC 239, 1947-1953. 132. Con, C. F. (1956). In “Enzymes: Units of Biological Structure and Function” (0.H. Gaebler, ed.), pp. 573-583. Academic Press, New York. 133. Danforth, W. H., Helmreich, E., and Cori, C. F. (1962). PNAS 48, 1191-1 199. 134. Helmreich, E., and Cori, C. F. (1965). Adv. Enzyme Regul. 3, 91-107. 135. Lyon, J. B., and Porter, J. (1963). JBC238, 1-11. 136. Danforth, W. H., and Lyon, J. B. (1964). JBC 239, 4047-4050. 137. Rahim, 2. H. A., Perrett, D., and Griffiths, J . R. (1976). FEES Lerr. 69, 203-205. 138. Black, W. J., and Wang, J. H. (1968). JBC 243, 5892-5898.

Phosphorylase Kinase CHERYL A. PICKETT-GIES DONAL A. WALSH Depariment of Biological Chemistry School of Medicine Universiiy of California, Davis Davis, California 95616

I. Introduction ............................................. 11. Physicochemical Properties A. Sources and Isolat

B. Subunit Structure ................... C. Isozymes . . . . . . D. Spatial Arrangement of Subunits ................................ E. Subunit Isolation . . . . . . . . . . . . . . . . . . F. In Vivo Subunit Synthesis ..................... 111. Subunit Function and Interaction between Subunits .............. A. The Catalytic Subunit(s) . B. The &-Subunit and Nature of Its Interactions in the Holoenzyme . . . . . . C. The a- and P-Subunits as Regulators . . . . . . . . . . . . . . . . . .

399 401 403

406

C. Peptide Substrate Specificity D. Nucleotide Substrate Specifici

B. Mg2+ and MgZ+-Ca2+ Interactions ............................ C. Regulation by Extrinsic Calmodulin and Troponin C . . . . . . . . . . . . . . . . D. Regulation by Glycogen and in the Presence of the Constituents of “Glycogen Particle” ....................................... VI. Proteolytic Activation of Phosphorylase Kinase .......................

419 422 425 428

395 THE ENZYMES, Vol. XVII Copyright Q 1986 by Academic Press, Inc. All rights of reproduction in any form reserved.

396

CHERYL A. PICKETT-GIES AND DONAL WALSH

VII. Covalent Regulation: In Virro Studies . . . . . . . . . . . . . . A. Intrinsic Phosphate Content .................................... B. Phosphorylation: General 'Considerations C. Cyclic AMP-Dependent Protein Kinase-C D. Autophosphorylation ...................................... E. Phosphorylation and Activation by Other Protein Kinases . . . . . . . . . . . . F. Phosphorylase Kinase D lation ......................... G. ADP-Ribosylat VIII. Regulation of Phos .................... ............... A. Hormonal Acti Phosphorylation in Intact Tissues ................................ References . . . ..... .......

429 429 430 43 I 441

444 445 446 446 446 450 453

1. Introduction Protein phosphorylation and dephosphorylationare well recognized as a major metabolic control mechanism in eucaryotic cells. In mammalian systems, phosphorylation plays a role in the regulation of protein function in such diverse processes as carbohydrate and lipid metabolism, gene regulation and macromolecular synthesis, membrane transport, ionic homeostasis, muscle contraction, cytoskeletal organization, neural transmission, oncogenesis, and many others. Many avenues that have led to our understanding of regulation by protein phosphorylation have been obtained from studies of liver and muscle glycogen metabolism, and this area of investigation has often served as the initiation point for the elucidation of the primary principles that govern such regulatory phenomena. It was during studies of muscle glycogenolysis that the phosphorylation of glycogen phosphorylase, the first example of enzyme regulation by phosphorylation, was discovered. By the 1940s, through work in the laboratories of Con and others, the basic biochemistry of glycogen breakdown had been established with the characterization of the phosphorylase-catalyzed conversion of glycogen to glucose 1-phosphate. At the same time, it was well known that exposure of liver cells to epinephrine caused a rapid breakdown of glycogen and an increase in free glucose, suggesting that epinephrine was somehow acting to increase the activity of phosphorylase (1). During the 1940s, workers in the Cori laboratory (2) demonstrated that phosphorylase existed in two molecular forms, but the nature of the difference and the mechanism by which these forms were interconverted remained unclear for more than a decade. Soon after the first protein kinase was described in 1954 ( 3 ) ,it was established that the two forms of phosphorylase were interconverted by phosphorylation-dephosphorylation (4). In 1950, Sutherland ( 1 ) demonstrated that epinephrine elicited the production of a heat-stable factor that was capable of activating phosphorylase in cell homogenates. This factor was soon identified as adenosine 3',5'-monophosphate or CAMP(3,and was shown to be produced from ATP by the action of a mem-

10. PHOSPHORYLASE KINASE

397

brane-bound adenylate cyclase (6). Meanwhile, characterization of the kinase responsible for the phosphorylation of phosphorylase was initiated in the laboratory of E. H. Fischer and E. G. Krebs. It became clear that this enzyme was also subject to phosphorylation and this modification increased its catalytic activity toward phosphorylase (7,8). Addition of CAMP to purified phosphorylase kinase in the presence of Mg2 and ATP stimulated the activation of the enzyme, thus the ground work was laid for the hypothesis of a “cascade” system in the hormonal regulation of glycogen metabolism. For a time it was thought that this sequence of events was complete; however, further investigation demonstrated that CAMPdid not act directly on phosphorylase kinase but rather upon another enzyme contaminating the preparations. Identification of this enzyme, the CAMP-dependent protein kinase, by Walsh et al. (9) in 1968, completed the basic sequence of steps in the f3-adrenergic regulation of phosphorylase. Continuing studies on the enzymes of glycogenolysis have upheld the basic scheme of events as initially presented for the regulation of glycogen breakdown, but equally important, have shown that the system is much more complex than originally envisioned. From the studies of these enzymes, and of other enzymes that have since been shown to be regulated by protein phosphorylation, it has become abundantly clear that regulation by protein phosphorylation can often involve highly intricate interrelationships of multiple protein phosphorylation control mechanisms. Some of these complexities include the following: +

1. Protein phosphorylation may either activate or inactivate the biological

function of a protein. 2. Changes in activity may be reflected by alterations in the affinity for substrates, the affinity for allosteric ligands, the maximum velocity of the reaction, or combinations of some or all of these. 3. Protein phosphorylation can occur in multiple sites on a single polypeptide chain, catalyzed either by the same or by different enzymes. 4. The status of phosphorylation of a site can regulate not only the activity of the protein, but also the regulation by phosphorylation of other sites. 5 . The system for protein dephosphorylation is as deeply complex as the system for phosphorylation. Protein phosphatases are, like the protein kinases, subject to a variety of regulations including protein phosphorylation and dephosphorylation. 6. Specific regulation by the phosphatases and kinases does not appear to follow a simple pattern. Different sites phosphorylated by the same kinase might require different phosphatases for their dephosphorylation and, vice versa, the same phosphatase can catalyze dephosphorylation of sites phosphorylated by different kinases. The above are but a few of the complexities that have become apparent in systems regulated by protein phosphorylation; in fact, from the simple schemes of the glycogenolytic cascade presented in the 1960s and early 1970s, which

398

CHERYL A. PICKE'IT-GIES AND DONAL WALSH

appeared to adequately describe the physiological manifestation of such regulation, we have reached a stage in which it is very difficult to put into a physiological context the complexities of the processes that are now apparent. It is within this framework that this chapter on phosphorylase kinase is presented. Here, we attempt to at least indicate the extensive network of regulatory phenomena that impinge upon the control of phosphorylase kinase activity. For many, however, a full physiological rationale cannot be completely presented. While we have attempted to be complete, not all of the information that is available about phosphorylase kinase can be given adequate treatment; such can only be obtained from the cited original work. Three notable reviews on phosphorylase kinase have been presented by Chan and Graves (ZO), Carlson et af. ( I ] ) , and Malencik and Fischer (12). The understanding of the control of glycogenolysis has served well for many years as a guideline for how other metabolic control systems may function. Similarly, an understanding of phosphorylase kinase can serve as a model to understand the properties of other regulatory enzymes. An attempt is made in this chapter to provide this understanding.

II. Physicochemical Properties A. SOURCES AND ISOLATION Phosphorylase kinase was first isolated from rabbit skeletal muscle (7) where it constitutes approximately 0.5% of the soluble protein. The commonly employed purification procedures involve the sequential use of (1) acid precipitation of the glycogen pellet and associated enzymes (including phosphorylase kinase) from crude extracts and/or, (2) two-step differential ultracentrifugation, (3) ammonium sulfate precipitation, and (4) gel filtration (13, 14). In addition to these procedures or as alternatives to various steps, several investigators have utilized purification techniques such as DEAE chromatography (14), affinity chromatography on immobilized calmodulin (15) or phosphorylase (16), hydrophobic chromatography (17), and sucrose density-gradient ultracentrifugation (18). Using a variety of these methods, the enzyme can be obtained in an essentially homogeneous form, although often the only criterion of homogeneity applied has been SDS-polyacrylamidegel electrophoresis (SDS-PAGE). Minor contaminants (- 1%) are frequently present; whether such contaminants might have affected the outcome of results has often been ignored, perhaps inappropriately. In addition to the rabbit skeletal muscle enzyme, phosphorylase kinase has been purified from mouse (19), dogfish (20), and red bovine (21) skeletal muscle; from bovine cardiac muscle (22); from chicken gizzard (22a); and from rat liver (23). All of these enzymes appear to share many very similar characteristics with respect to physical properties, enzymic activities, and modes


399

of regulation. Since by far the predominant number of in vitro studies have utilized the rabbit skeletal muscle enzyme, these results constitute most of what is reported in this review. Where there are established differences for the enzyme from other sources, they are noted. One phosphorylase kinase that appears to be quite distinct in molecular size, subunit composition, and enzymic characteristics, is that obtained from yeast (24); the studies so far on this enzyme, however, have been quite limited (24, 25). B. SUBUNIT STRUCTURE The isolated, homogeneous rabbit skeletal-muscle phosphorylase kinase has a molecular weight of approximately 1.3 X lo6 (13, 14) and is composed of four types of subunits in the stoichiometry a (a’)4p4y484.Two isozymes have been identified that differ in the size of the largest subunit, designated a or a’. Molecular weights of these subunits, based on their electrophoretic mobility in SDS-polyacrylamide gels, have been reported as a = 145,000 (14), 136,000 (26), 127,000 ( 2 3 , 118,000 (13); a‘ = 133,000 (26), 134,000 (22), 140,000 (14);p = 128,000 ( 1 4 , 120,000 (26), 113,000 ( 2 3 , 108,000 (13);with the first listed ones being the most generally accepted. Absolute values for y = 44,673 (28) and 6 = 16,680 (29) have been determined from sequence analysis and are in primary agreement with prior values determined by SDS-PAGE. C.

ISOZYMES

When first isolated from mixed muscle types the minor a‘ band seen in SDSgels was suspected to be a proteolytic product from the more abundant a band. However, further purification or treatment of extracts to inhibit proteolysis did not alter the relative amount of a’observed. Jennissen and Heilmeyer (26) and later Burchell et al. (30) prepared phosphorylase kinase from red (slow-twitch) and white (fast-twitch) skeletal muscle. Characterization of the enzyme suggested that the a‘ enzyme was predominant in slow-twitch muscle whereas the a enzyme was predominant in fast-twitch muscle. This evidence, coupled with the lack of a subunit in purified cardiac phosphorylase kinase ( 2 2 ) , led to the generally accepted hypothesis that two isozyme forms of phosphorylase kinase exist, one form containing four a-subunits and the other four a‘-subunits, and that each isozyme is specific for a particular type of muscle fiber. Separation of these isozymes appears to be best accomplished by calmodulin-affinity chromatography (15). There are several questions about phosphorylase kinase isozymes that need to be addressed. Phosphorylase kinase exists in most nonmuscle tissues but for only one of these, liver, has the isozyme form been evaluated. The isozyme that has been purified from liver appears most likely to be the a type ( 2 3 ) ,but this protein

400


was purified starting from the proteins associated with the glycogen complex, which for liver phosphorylase kinase represents only 5% of the total enzyme present in the cell. Whether this isozyme also constitutes the bulk of phosphorylase kinase present in liver is not known. (Studies of this type have been severely hampered by problems of proteolysis during purification, to which the phosphorylase kinase appears particularly susceptible. Often even extensive use of protease inhibitors does not negate the problem.) It is not known for any tissues whether within a single cell type one or both isozymes are expressed andlor whether heterologous enzyme containing both a-and a’-subunits in the same molecule can occur. Whether the two isozymes of phosphorylase kinase have unique physiological functions is not known. While this would appear most reasonable and there are clear differences in in vitro properties, it is not apparent what different physiological purposes the two isozymes might serve. The most plausible, albeit speculative, proposal is that each is differentially regulated in response to Ca2+,this is discussed in more detail in Sections V,A and VII1,A. The original correlation (26, 30) between phosphorylase kinase isozyme-type and muscle fiber contractile property (i.e., fast- or slow-twitch) should be expanded. Rat skeletalflexor digitorum brevis muscle,which is 90% FOG fibers (fast-twitch oxidative glycolytic), contains almost entirely the a’isozyme (30a),and high levels of a’ isozyme have also been reported in other fast-twitch muscles containing high proportions of FOG fibers (30b). Thus phosphorylase kinase isozyme distribution in muscle is best related to metabolic rather than contractile properties with the a isozyme being present in cells that rely primarily on glycolytic activity [FG (fast-twitch glycolytic) fibers] and the a’ isozyme in tissues with higher oxidative capacity [FOG and SO (slow-twitch oxidative) fibers and cardiac muscle]. Recently, Lawrence et al. (30b) have reported that repetitive long-term stimulation (- 10 weeks) of muscle prompted a change in isozyme distribution with an elevation of the a’isozyme. This could well correspond to an increased proportional mass of oxidative fiber types. It has been suggested that differences in phosphorylase kinase forms (isozymes) may also be related to their subcellular distribution. Studies by Schwartz et al. (3J),Horl and Heilmeyer (32),and Le Peuch et al. (33)have indicated that an endogenous “phosphorylase kinase” was associated with the sarcoplasmic reticulum, that added phosphorylase kinase modulated sarcoplasmic reticulum Ca2+ transport, and that this appeared to be due, at least in part, to the phosphorylation of the protein ‘‘phospholamban. The substrate specificity of this sarcoplasmic reticulum phosphorylase kinase and the standard enzyme isolated from cytosol (and glycogen particle) are apparently identical but the two enzymes appear to be distinct based upon immunogenicity, the requirement for calmodulin for activity, inhibition by fluphenazine, and the retention of the sarcoplasmic reticulum enzyme in I-strain mice in which the cytosol-glycogen ”


40 1

particle enzyme is genetically absent. The relationship between these two enzyme forms, in terms of structure, subunit composition, etc., has not been evaluated. Most likely, the membrane associated enzyme is not truly a "phosphorylase kinase. " Multiple forms of phosphorylase kinase might also exist based upon size and possibly subunit stoichiometry. In one of the original studies of the purification of phosphorylase kinase (13), it was recognized that the holoenzyme (M,= 1.3 X lo6) tended to aggregate with the formation of complexes that were large enough to result in turbidity. Upon further characterization (12), this turbid fraction appeared to be rather contaminated by another polypeptide, suggested possibly to be a-actinin. This turbid phosphorylase kinase fraction has generally been regarded to be artifactual, arising as a consequence of pressure-dependent aggregation. Subsequently, Hessova et al. (34) observed that the higher-molecular-weight fraction also occurs as a consequence of heparin treatment. Of interest, heparin treatment promotes the formation of the phosphorylase kinase activity that these authors designate A, (see Section V,B), a form that is calciumindependent and that they suggest plays an important role in the cell in permitting Ca2 -independent phosphorylase a formation. Concomitant with a decreased sensitivity to Ca2+, this turbid fraction appears to contain a decreased level of the &subunit (34). Determination of the stoichiometry of phosphorylase kinase does have some inherent difficulties but the possibility that the enzyme might exist in other than the a (or p4y4ti4form definitely merits further consideration (12, 22). The actual data demonstrating a precise subunit stoichiometry is neither extensive nor definitive, especially for the documentation of the ratio of y to either a or p. +

D.

SPATIAL

ARRANGEMENT OF SUBUNITS

Clues to the potential spatial arrangement of phosphorylase kinase have come from electron microscopy (35, 35a), cross-linking studies (36-39), and isolation of partial structures (40). By either conventional negative staining or scanning transmission EM (35a), phosphorylase kinase appears as a bilobal structure resembling two opposing parentheses held together by two short cross bridges. This butterfly-like nonglobular structure has the molecular mass of the hexadecamer holoenzyme. Selective chymotryptic cleavage of the a subunit did not destroy the overall structure except that the individual lobules were misshapen. Restricted tryptic cleavage of both a and p subunits tended to convert the molecule to two equal halves, suggesting that the native molecule is comprised of bridged octamers each of structure a2p2y2G,. Interestingly, even with tryptic cleavage of a high percentage of a and p subunits, an unexpected proportion of molecules remained in the butterfly structure of the original mass of the holoenzyme. These patterns of effects suggest that both the p subunit and strong

402


noncovalent forces contribute to the general conformation of the cross bridges in the native molecule. Earlier studies have also provided important insights into subunit-subunit interactions. Chan and Graves (43, using lithium bromidepromoted dissociation, obtained two partial complexes, a y S and yS, both of which retained catalytic activity; the characteristics of expression of activity would suggest that these two complexes retained some (or most) of the same interactions between the subunits as exist in the native molecule. From crosslinking studies, using dimethyl suberimidate, dimeric pairs of act, pp, ay, and Py, and trimers of af3y and py2 have been obtained (36-33, indicating close interactions of these various combinations of subunits in the native molecule. Also of interest, similar cross-linking approaches by Fitzgerald and Carlson (38) using difluorodinitrobenzenehave shown that on enzyme activation by a variety of means an increased P-P interaction is observed. The current 3-dimensional structure of phosphorylase kinase is unknown; whatever eventual structure the holoenzyme proves to have, it will need to account for the various observations listed above. One early schematic presented by Picton et ul. (39) to account for most of the cross-linking observations and the general “butterfly” EM observations is depicted in Fig. I, but this does not readily account for the most recent EM observations (3%). Also shown in the schematic model is the interaction of four additional calmodulin molecules ( S ‘ ) which by cross-linking studies (37) have been shown to be associated with a- and p-subunits. The role of these additional calmodulin interactions is discussed in Section V,C). E. SUBUNITISOLATION The individual subunits a,a’,and p have yet to be isolated in a form that maintains their full structural integrity. The a-,p-, and y-subunits can be iso-

Activation

Activotion

J FIG. 1. Pictoral arrangement of the subunits of phosphorylase kinase. The arrows depict one type of conformational change suggested to occur upon activation. Adapted from Picton ef al. (39).


403

lated in a denatured form by high-performance liquid chromatography (HPLC) (27) and the y-subunit also by Sephadex G-200 gel filtration in the presence of SDS (13). Initially, the y-subunit was reported to have been isolated with retained catalytic activity (41), however, the form isolated was subsequently identified as the y8 dimer. Pretreatment of the yS dimer with EGTA, followed by sucrose density-gradient ultracentrifugation resulted in the formation of a Ca2 independent kinase activity which, although not specifically tested, may well be an isolated y-subunit with full structural integrity (40). The y-subunit, isolated by HPLC, has been obtained following renaturation, in an active form (42). The &subunit can be readily isolated with retained function by heat denaturation of the holoenzyme (43). +

F. In Vivo SUBUNIT SYNTHESIS The synthesis of subunits a and P from y in vivo has been reported (43a). In that study the turnover of the a and P subunits in skeletal muscle appeared to be -1.6 fold greater than that of the y subunit. Clearly more data of this type is needed.

111.

Subunit Function and Interaction between Subunits

A. THECATALYTIC SUBUNIT(S)

The function of the individual subunits of phosphorylase kinase is only partially understood. Several reports indicate that the y-subunit has catalytic activity. This was first proposed by Hayakawa et al. (44) when it was observed that activation of the enzyme by trypsin treatment occurred with proteolysis of the aand P-subunits, whereas the y-subunit remained intact. More direct evidence has come from the work of Skuster et al. ( 4 4 , Chan and Graves (40, 45, 46), and Kee and Graves (42) who have isolated a y8 complex that retained full catalytic activity for the phosphorylation of both phosphorylase and phosphorylase kinase. Furthermore, treatment of the y8 complex with EGTA followed by ultracentrifugation led to a Ca2 -independent phosphorylase kinase activity, but Ca2 -dependency could be restored with addition of exogenous &subunit (46). Equally strong evidence that the y-subunit contains a catalytic site comes from the recognition that there is substantial homology between the sequence of the ysubunit and that of the catalytic subunit of the CAMP-dependent protein kinase +

+

(28).

There is also evidence, albeit less direct, that the P-subunit might contain another catalytic site (distinct from that on the y-subunit). This is supported by several lines of experimentation. One approach has been the use of ATP analogs

404


to covalently label ATP binding sites on phosphorylase kinase. Gulyaeva er al. ( 4 3 , utilizing alkylating ATP analogs modified in the triphosphate moiety, found that both P- and y-subunits were labeled. Modification of the @-subunit correlated well with inactivation of the enzyme while y-subunit modification appeared to have little effect on activity. Subsequently, King et al. (48, 48a), demonstrated preferential labeling of the P-subunits of phosphorylase kinase with two photoaffinity analogs of ATP, 8-azido ATP and its 2’,3’-dialdehyde derivative, both of which serve as the phosphoryl-donor substrate and thus must interact at the catalytic site. With either of these, labeling of the P subunit was accompanied with loss of activity, but correlations were inexact and greater (faster) derivatization occurred than inactivation. Both labeling of the p subunit and inactivation (on a percentage basis) were equally protected by ADP; there were, however, disparate effects with addition of divalent cations in that labeling was depressed in the presence of Mg2+ or Ca2+, whereas inactivation was unaffected by Mg2+ and enhanced by Ca2+. Affinity labeling has also been studied by this group using 5 ’-p-fluorosulfonyl benzoyl adenosine (48b);again, the p subunit was preferentially labeled and the enzyme inactivated, but because other subunits (aand y) were also derivatized, no conclusions could be reached about exact correlations. One difficulty in interpreting such data is that phosphorylase kinase has been shown to have eight binding sites for ADP (see Section IV,A) and this leaves open two possibilities. The first is that there are four catalytic sites (one on each y) and four allosteric sites (on p?); the second is that the latter are also catalytic sites. ADP, however, is clearly an allosteric regulator. The presence of a catalytic site on the P-subunit has also been suggested by Fischer et al. (49), who reported isolation of a catalytically active phosphoprotein after proteolysis of phosphorylase kinase phosphorylated predominantly in the p-subunit; a full report of these findings has not been presented. Compatible with this observation, however, Killilea and Ky (50) observed that following extended trypsin treatment of cardiac phosphorylase kinase, only one polypeptide remained, which corresponded to the P-subunit, but catalytic activity had been retained. Similar indirect evidence also exists suggesting that the a subunit might contain a catalytic site. Crabb and Heilmeyer (27) have shown that there is some sequence homology between the N-terminal region of the a-subunit and that of the transforming protein from Rous sarcoma virus, which is a tyrosine protein kinase. ATP-dependent derivatization of the ci subunit has also been reported using fluorescein isothiocyanate (50a), which with several proteins binds at or near an ATP catalytic site. Very selective derivatization of the a (and a’)subunit was seen that could be blocked by ATP addition (50a, 50b), but as with the data described above for the p subunit this could possibly represent binding to a regulatory rather than a catalytic site. The sequence of the FITC derivatization site has been reported (3%). Evidence for potential multiple catalytic sites also stems from studies of phos-

405


phorylase kinase-catalyzed phosphorylation of substrates other than phosphorylase b. Carlson and Graves (51) suggested the possible existence of two separate catalytic sites to explain the enhanced autophosphorylation they observed when autocatalytic reactions were carried out in the presence of phosphorylase or peptide analogs of its phosphorylated region. They also observed that troponin did not inhibit the phosphorylase b to a conversion at pH 8.2, and in fact, accelerated the reaction at pH 6.8. One would normally predict that addition of one substrate (phosphorylase in the former case, troponin in the latter) would competitively inhibit the phosphorylation of another substrate (phosphorylase kinase and phosphorylase b, respectively). Dickneite et al. (52) reported that antibodies to phosphorylase kinase inhibited phosphorylase b phosphorylation in an uncompetitive manner but inhibited troponin phosphorylation in a competitive manner. This type of differential inhibition of activity toward various substrates was later described by King and Carlson (53)using an ATP analog to affinitylabel phosphorylase kinase. Affinity-labeling in the presence or absence of Ca2 and Mg2+ allowed them to distinguish three different classes of substrates based on their reactivity with the partially inactivated enzyme. In explaining their results, the authors proposed a model in which glycogen synthase and phosphorylase b are preferentially phosphorylated at one type of catalytic site, whereas troponin I and troponin T are phosphorylated at another. Further evidence supporting the idea of multiple catalytic sites has been reported by Kilimann and Heilmeyer (54, 55) who have distinguished three separate activities of phosphorylase kinase towards phosphorylase b by their dependence on Ca2 ,Mg2 , NH,Cl, and pH. These three types of enzymic activity have different apparent specificities toward the protein substrates phosphorylase b, troponin I and T, and phosphorylase kinase; albeit that the proposed different substrate specificities from the two reports (53-55) do not match. Although the several lines of evidence presented here suggest that the ysubunit and another subunit, potentially p, both contain catalytic sites, the data are still ambiguous. There can be little doubt, especially with the work of Chan and Graves (40, 45, 46) and Reimann et al. (28), that the y-subunit has a catalytic site. However, many of the observations suggesting that there is a second site can be rationalized if one assumes complex interactions within the phosphorylase kinase molecule. That phosphorylase kinase might exhibit complex interactions would hardly be surprising in the light of what is already known concerning its regulation. Chan and Graves (45) have potentially provided the most important clue that it is quite likely only the y-subunit that contains a catalytic site (at least for phosphorylase 6 ) . They have shown that the molar activities of the holoenzyme, the ay6 complex and the y6 complex (plus additional calmodulin) are, respectively, 99.3, 91.4, and 104 molecules/sec with phosphorylase as substrate. Thus, the y6 complex (devoid of a and p) and the ay6 complex (devoid of p) exhibit the full catalytic competence of the holo+

+

+

406

CHERYL A. PICKETT-GIES AND W N A L WALSH

enzyme. This would obviate the need to evoke a second catalytic site, in particular for the phosphorylase b to a conversion. However, other substrates might be phosphorylated by other catalytic sites, and it has been reported that phosphorylase kinase exhibits a low level of phosphatidyl inositol kinase activity which constitutes quite a different type of substrate (55a). Also of interest in a study of monoclonal antibodies directed against phosphorylase kinase, one clonal antibody was found that interacted equally with the a,p, and y subunits (55b). If each contained a similar catalytic site to which the antibody was directed, this may be as would be expected.

B. THE SUBUNIT AND NATUREOF ITS INTERACTIONS IN THE HOLOENZYME The existence of the &subunit of phosphorylase kinase was not demonstrated until 1978 (43) due in large part to its small size and poor staining with typical protein stains. The identity of the &subunit and calmodulin, a calcium-binding protein first identified in the brain, was suggested by several physicochemical properties and was confirmed by its amino acid composition and ability to reactivate calmodulin-dependentenzymes (43). The amino acid sequence of the b-subunit has been found to be identical to that of bovine uterus calmodulin and to differ only in amide assignments at two residues from that of bovine brain calmodulin (56). Calmodulin acts as a Ca2+-dependent modulator of a wide variety of enzymes and the mediator of the control of these enzymes in response to physiological fluxes of Ca2+. If Ca2 binding to the holoenzyme occurs exclusively through the &subunit, one might expect that phosphorylase kinase would show similar binding properties to those observed with calmodulin; any observed differences may give an indication of restrictions placed upon the &subunit as a consequence of it being an integral component of the holoenzyme. This analysis has been made by Heilmeyer et al. (57-59) who presented a comparison of the Ca2+-binding properties of the holoenzyme and its isolated &subunit. Their data are summarized briefly in a simplified form in Table I. Analyses of binding were performed under three conditions: low ionic strength, high ionic strength, and high ionic strength plus Mg2+. At low ionic strength, the holoenzyme binds 3-4 mol of Ca2+ per aPy8 with high affinity (Kd= 20-1000 nM). At high ionic strength, in the absence of Mg2+, two of these high-affinity sites are retained but two are lost. In the presence of Mg2+, two effects occur-the affinity of the two retained sites is diminished, but two other high-affinity sites are now detectable. Kohse and Heilmeyer (59) classified the sites as Ca2 -Mg2 and Ca2 -specific. The Ca2+/Mg2+ sites bind either ion so that in the presence of Mg2+ the apparent affinity for Ca2+ is depressed. The Ca2 -specific sites bind only Ca2 +

+

+

+

+

+

407


TABLE I COMPARISON

OF

CA2+ BINDING TO CALMODULIN OR PHOSPHORYLASE KINASE~

n

Kd

(M)

nb ~~

b

X

OF

Site

Kd(M) ~

3 1 High ionic strength 2 4.0 X 2 2 4.0 x High ionic strength 2 6.6 X 2 plus Mg2+ 2 2.8 X loWs 2

Low ionic strength 4 5.8

&SUBUNIT

Phosphorylase kinase

Isolated calmodulin Conditions

THE

]OW8

2.0 x 10-8 6.0 x 10-6 2.0 x 10-8

I

Ca2+/Mg2+ plus Caz+-specific Ca2+IMg2+ CaZ+-specific 2.5 x lo-’ Caz+/Mg2+ 3.0 x 10-6 CaZ+-specific

Adapted from Kohse and Heilmeyer (59). Per (apyS).

but, in the case of the holoenzyme at high ionic strength, the Ca2 -specific sites require the presence of Mg2+ for Ca2+ binding. The designations for these sites are indicated in Table I. With the isolated &-subunit,at low ionic strength, 4 mol of Ca2+ are bound per mol with high affinity, with both stoichiometry and affinity similar to what is observed with holoenzyme. At high ionic strength, the isolated &-subunitretains four Ca2 -binding sites, but their affinity is reduced. Addition of Mg2+ reduces the affinity of two of these sites (the Ca2+-Mg2+ sites) further but does not affect binding to the Ca2+-specific sites. Cooperativity with Hill values of -2 was obtained for Ca2+ binding to both types of sites in both the isolated &-subunit and holoenzyme. From these results, Kohse and Heilmeyer (59) drew the following conclusions: +

+

1. The similarities of binding of Ca2 to the isolated &subunits and holoenzyme, especially with respect to number of sites, designation of types of sites, and the cooperativity of Ca2 binding, indicates that Ca2+-binding by phosphorylase kinase can be fully accounted for as occurring through the &subunits. 2. The fact that only in isolated &-subunits is the affinity of Ca2+ for the Ca2 -Mg2 sites depressed by an increase in ionic strength suggests that the integration of calmodulin into the holophosphorylasekinase stabilizes it in a conformation that is similar to that of the isolated subunit at low ionic strength. 3. Subunit-subunit interaction in the holoenzyme, most likely involving heterologous subunits, modifies the conformation of the environment of the Ca2 -specific sites so that binding occurs only in the presence of Mg2 . +

+

+

+

+

+

408


The specific Mg2+-binding sites may be on the b-subunit or elsewhere on the phosphorylase kinase molecule. The interaction of the b-subunit with the other subunits of phosphorylase kinase differs from that of calmodulin with other calmodulin-regulated enzymes. As exemplified by the regulation of myosin light chain kinase by calmodulin and presented in Chapter 4 in this volume (60),the typical mode of calmodulin interaction can be described by the two-step reaction as shown in Scheme I. Calmodulin

+ Ca2+ + Ca2+ -calmodulin

enzyme

+ Ca2+ -calmodulin-enzyme

SCHEMEI

That is, calmodulin binds only in the presence of Ca2+, and when Ca2+ is removed, calmodulin dissociates from the enzyme. In contrast to this, the bsubunits are tightly bound integral components of phosphorylase kinase holoenzyme, and are not readily dissociated by such agents as the Ca2+-chelators, EDTA and EGTA, or by high concentrations of urea (12,37,43).[A slow rate of exchange (15% per week) of I4C-labeled exogenous calmodulin with the 6subunit can occur (37)l. Cross-linking studies (36, 37) have indicated that the 6subunit is primarily bound to y-subunit. This interaction is maintained during the lithium bromide-promoted dissociation to form the partial complexes a y b and yb (40);however, it appears that at the level of the y6 dimer the interaction is more closely analogous to other calmodulin-regulated enzymes since the dimer can apparently be dissociated by treatment with EGTA (46). C. THEa- AND P-SUBUNITS AS REGULATORS In addition to the possible functions previously discussed, several lines of evidence suggest that the a- and P-subunits serve a regulatory function. Phosphorylation of the a-and P-subunits by the CAMP-dependent protein kinase (13, 14) or by autophosphorylation (61, 62) results in activation, as does limited proteolytic degradation of these subunits (14, 63). Activation of the enzyme also results from dissociation of the holoenzyme by LiBr (45), which has led to the suggestion that the activity of the enzyme is inhibited by the regulatory subunit(s) (aand p), and that this inhibition can be relieved by phosphorylation, limited proteolysis, or dissociation. In consideration of potential roles of phosphorylase kinase subunits, a possible indicator of unique function is the observation that whereas, for those enzymes tested, the p-, y- (and 6-) subunits appear identical (by SDS-PAGE), this is not true for the a-subunit. Not only is it clear that within species there are two forms of a-subunit (a and a’)but between species, the primary subunit that appears different is the a (or a’) (64).Two recent reports have described the preparation


409

of subunit specific antibodies which promise to be of some assistance in further defining the specific roles that each subunit serves (55b, Ma). D.

DOES THE

y-SUBUNIT HAVEADDITIONAL ROLES?

In a single report by Fischer et al. (49), it was indicated that there was considerable homology between the y-subunit of dogfish phosphorylase kinase and dogfish actin, even to the extent of interaction with myosin. None of this, however, was commented on in the subsequent follow-up full-length paper (20) and there is clearly no homology between equivalent rabbit muscle proteins (28). The reason for the high propensity of phosphorylase kinase to aggregate (13) (it does even with pressure which normally promotes disaggregation) is not known, and a similarity of the y-subunit with actin would be an attractive explanation. The polymerized form has been reported to contain another component of the contractile apparatus, a-actinin ( 2 2 ) .

IV. Catalytic Properties A.

CHARACTERISTICS OF THE PHOSPHORYLASE b TO a REACTION

The major reaction thought to be catalyzed by phosphorylase kinase in vivo is the phosphorylation of phosphorylase b. In this reaction, phosphorylase b, a dimer, is phosphorylated at each of two identical serine residues in the presence of Mg2 and ATP. Mg2 added in excess of that required to form the substrate, MgATP2-, results in stimulation of phosphorylase kinase (7, 65, 66). This may be via additional binding sites for free Mg2+ ( 6 3 , although free A T P - has been suggested to be inhibitory and thus some uncertainty exists as to whether free Mg2+ is stimulatory or free ATP4- inhibitory (7, 68, 69). In addition to Mg2 , Ca2 is required for the activity of both the activated and nonactivated form of the enzyme (7, 70). The allosteric effects of Ca2+ and Mg2+ are discussed in more detail in Section V,B. Cheng ef al. (71) have studied the interaction of ADP with phosphorylase kinase. In addition to being a product, ADP is an allosteric activator; 8 mol are bound per (aPy13)~ with Kd values in the range of 0.26 to 17 pM. ADP stimulates both phosphorylase conversion and autophosphorylation and inhibits @-subunitdephosphorylation. Binding at this allosteric site is highly specific for the ADP moiety and many ADP analogs could not substitute for it (71). Nonactivated phosphorylase kinase isolated from resting muscle (in the presence of divalent cations and in the absence of phosphatase inhibitors) has little activity at physiological pH (pH 6.8-7.0), but has considerable activity at pH +

+

+

+

410


values greater than 7.6 (7). Following phosphorylation, catalyzed by one of several protein kinases, or following limited proteolysis, the activity at pH 6.87.0 increases markedly and much more so than that measured at higher pH values (e.g., pH 6.8 activity increases as much as 50-fold with phosphorylation by the CAMP-dependent protein kinase). These properties are maintained during purification of the protein to homogeneity. Since dissociation of the enzyme into partial complexes (ay8 and y8) also results in a marked activation at pH 6.8 with little change in activity at pH 8.2, it is most likely that the regulatory subunits (a and p) in nonactivated enzyme inhibit the catalytic site from exhibiting maximum catalytic potential; this inhibition is relieved either by a conformational change induced by pH or covalent modification, or by removal of the inhibitory subunits by dissociation and/or proteolysis. The changes in activity that occur either with pH or covalent modification can be attributed almost entirely to changes in affinity for phosphorylase. This change in pH dependency has been exploited as a means to express the activation status of phosphorylase kinase, especially for an evaluation of enzyme activation occurring in intact tissues. Thus, nonactivated phosphorylase kinase has a ratio of activity at pH 6.8 to that at 8.2 of -0.04-0.08, and activation by pH change, covalent modification, or proteolysis increases the activity ratio to -0.2-0.9. Although this measurement has gained wide acceptance, there are some inherent problems with its use. For example, in studies of phosphorylase kinase activation in guinea pig hearts, Hayes and Mayer (72) could detect no changes in the ratio of activity at pH 6.8 to that at 8.2, despite a readily observable change in the pH 6.8 specific activity that was clearly a consequence of CAMP-dependent activation. The differences were shown to be attributable to differences between the kinetic parameters of guinea pig cardiac phosphorylase kinase and those of either the rat cardiac or rabbit skeletal muscle enzymes (72). It is important to note that the activity at pH 8.2 is not static but also changes with phosphorylation and proteolysis albeit, in most cases, less dramatically than the activity at pH 6.8 (44). Several endogenous factors also affect the pH 8.2 activity measurement and a greater variation in its quantitative value is often experienced (73). Some caution is therefore necessary in the interpretation of pH 6.8-8.2 activity ratios, and, we have repeatedly found that the measurement of pH 6.8 specific activity gives a more reliable index of phosphorylase kinase activation state.

B. KINETICS Kinetic studies of phosphorylase kinase have been difficult because of an unusual lag in its catalytic reaction. This lag is pH-dependent, being more It also marked at pH values near neutrality but not so apparent at pH 8.2 (I2,66). appears to be dependent upon buffers, preincubation with substrates, and enzyme

41 1


concentration. Several initial observations suggested that the lag may have been due to autophosphorylationof phosphorylase kinase (discussed in Section VI, D) with consequent activation (62). Subsequently, King and Carlson (74, 75) reported that the lag seen in phosphorylase conversion (and in autophosphorylation) at pH 6.8 can be diminished by preincubating the kinase with Mg2+ and Ca2 . This process, which was termed “synergistic activation,” requires the presence of both ions at half-maximal concentrations of 5 pM Ca2+ and 4 mM Mg2+. Activation, which is maximal in 2 min, is reversed by chelators and decreased by both ATP and phosphorylase b. King and Carlson (75) concluded that this synergistic activation by Ca2 and Mg2 is the primary cause of the lag in the phosphorylase kinase reaction and that autophosphorylation, if it occurs, is secondary. Presumably, Ca2+ and Mg2+ promote a slow conformational change in the phosphorylase kinase structure, a situation that has been termed “hysteresis” (66, 75). This Ca2+ plus Mg2 -dependent synergistic activation has also been shown to occur within the relatively physiological milieu of the glycogen particle ( 7 5 ~ ) . Table I1 presents a summary of reported kinetic constants for phosphorylase kinase (7, 45, 76); for simplicity, the data are presented in two parts, A and B, reflecting assays done by different laboratories. The trends in each are similar, and whether the apparent differences reflect minor differences in assay conditions or enzyme preparation is not known. The data in part A represent best what has been explored of the nature of activation by pH or phosphorylation; the data in part B are directed at what occurs upon dissociation of phosphorylase kinase into partial complexes. As previously noted, nonactivated enzyme has a pH 6.88.2 activity ratio of -0.05 and activation by phosphorylation results in a marked change in the pH 6.8 activity but minimal changes at pH 8.2. These activations, either by pH or covalent modification, are reflected in the kinetic constants (Table 11, part A). Between pH 7.0 and 8.5 there is a 10-fold decrease in the K,,, for phosphorylase with essentially no change in either the K,,,for ATP or the V,,, of the reaction. Similarly, phosphorylation of phosphorylase kinase decreases the K , for phosphorylase, at lower pH, without modification of the other kinetic parameters. Thus, activation of phosphorylase kinase, either by an increase in pH or by covalent modification, is attributable to a change in affinity for phosphorylase. The dissociation of phosphorylase kinase into the partial complexes ayS and yS occurs concomitantly with an increase in catalytic activity which is minimal at pH 8.2 but marked at pH 6.8 [i.e., the ratio of activity at pH 6.8 to that at 8.2 of holoenzyme, ayS, and yS are, respectively, 0.04-0.07, 0.50-0.60, and 0.91 .OO (45)]. These changes in activity upon dissociation were likewise reflected by changes in kinetic constants (Table 11, part B). As with increasing pH or covalent modification, activation by dissociation is accompanied by an increase in affinity for phosphorylase, reflected primarily at pH 6.8 rather than at pH 8.2. +

+

+

+

412

CHERYL A. PICKE'TT-GIES AND DONAL WALSH TABLE I1 KINETIC CONSTANTSFOR PHOSPHORYLASE KINASE Phosphorylase

ATP

Conditionso Enzyme

PH

A. Effects of activation 8.5 Nonactivated

Activated

8.2 1.6 1.4 7.0 8.2 1.5

K, (W) 33 40 125-250

7.0 B. Effects of dissociation Nonactivated 8.2 8.2 Activated 8.2 8.2 8.2 YS 8.2 6.8 a

310 17 31 20 250 230 80 110 91 94 83

Conditions

ATP Mg2+ K, Phos (d) (d)(d) (W)

3 3 3

10 10 10

Mg2+

( d l

10 10

33

10

0.24

33

10

0.38

33

10

0.22 0.26

100

10

-

10

(45) (45)

0.50

100

0.58 0.95 0.86

-

10 10 10 10

(45) (45) (45) (45)

NS 2.8

2.8

2.8 2.8 2.8

10 10 10 10 10 10 10

(8) (8) (8) (8) (76) (8) (8) (76)

0.31

NS 3 3

Ref.

100 100

NS, not specifically stated.

In the case of dissociation to the y6 dimer, there appears also to be a loss in affinity for ATP, presumably reflecting some role for the a and p subunits in the interactions of ATP with the holoenzyme. Kinetic studies with phosphorylase b as substrate are potentially subject to interpretative errors since effectors may be enzyme and/or substrate directed. For this reason, alternate substrates (which do not bind such factors as metal ions, nucleotides, buffers, or glycogen) have been sought. Tessmar and Graves (77) have studied a tetradecapeptide composed of the same amino acid sequence that surrounds the phosphorylated serine in phosphorylase. This peptide is phosphorylated at the same site as the native substrate. Also, the reaction with the peptide is similar to that with phosphorylase in several important aspects; the reaction shows the same type of lag in product formation, a similar pH dependence, and essentially the same Ca2+ and MgATP2- requirement. Although the K,,,for peptide is considerably greater than that for phosphorylase b (suggesting the involvement of a greater region of phosphorylase in binding to phosphorylase kinase than simply the 14 amino acids at the phosphorylation site or the require-


413

ment for a precise peptide chain conformation), once bound, the peptide is readily phosphorylated. Using this substance, Tabatabai and Graves (69) studied the kinetic mechanism of the phosphorylase kinase reaction. With activated phosphorylase kinase (phosphorylated by the CAMP-dependent protein kinase) and either the tetradecapeptide or phosphorylase b, initial rate data suggest a sequential-type mechanism. Competitive inhibition patterns with analogs of the tetradecapeptide or of ATP are consistent with a random bi bi mechanism. The reversibility of the phosphorylase b to a reaction (catalyzed by phosphorylase kinase) has also been studied. Early reports suggested that the reaction was irreversible (78) but later studies indicate that reversal can take place in the presence of glucose which tends to dissociate tetrameric phosphorylase a to a dimer (79). Interestingly, the latter report indicates that the pH dependence of the reverse reaction differs from that of the forward reaction and that phosphorylation of phosphorylase kinase by the CAMP-dependent protein kinase does not affect the rate of the reverse reaction.

C.

SPECIFICITY PEPTIDESUBSTRATE

In early studies of phosphorylase kinase (80) it was shown that a tetradecapeptide, isolated by chymotryptic digestion of phosphorylase and containing the seryl residue that was phosphorylated in the native molecule, could be readily phosphorylated by phosphorylase kinase, albeit with a fivefold lower V,,, and a fivefold higher K,. This peptide has served as the initiation point for studies of phosphorylase kinase substrate specificity (81-83). Phosphorylation of the peptide shares many of the characteristics of that of the native substrate. The reaction requites Ca2+, and exhibits a low pH 6.8-8.2 activity ratio with nonactivated phosphorylase kinase; the peptide is phosphorylated at a faster rate by enzyme activated either by phosphorylation or proteolysis, and the reaction shows the characteristic initial lag in reaction rate. Subsequently, it has been shown that glycogen synthase is also phosphorylated by phosphorylase kinase and peptides derived from it have also been examined as potential substrates (84,85). A summary of this data is presented in Fig. 2. Peptides 1-5 are based upon the sequence of glycogen synthase, peptides 6-34 on the sequence of phosphorylase. The phosphorylatable residue is Ser-7 in glycogen synthase and Ser-14 in phosphorylase. Peptide 5 is the sequence of the first fifteen residues of glycogen synthase; peptide 6 , the first eighteen residues of phosphorylase; and peptide 7, the originally identified tetradecapeptide (80). For comparison the two sequences derived from phosphorylase and glycogen synthase have been aligned for amino acid homology; the sequence of glycogen synthase has five residues deleted which would be approximately a little more than one turn of an (Y helix.

414


-, I

0 \

/

/Phosphorylasa Paptida~

\

.v

...............................

................................. .........,.................. .... Pro-Leu-%!?! ............................. Arq-Thr-Leu-Ser1 0 s 6 e r

.-k-1

Glyccqan Synlhase Paptides

L:u .Lv,'

I

$dl

I

Ap

K, Vmos Psptids rmMl Ipm0l/min/mgl X i

3.50 OBI

1.02

VoI- Ser-Ser-Leu-Pro-GI Leu Gln0.70 6 8 o lo t i la 1 2 - i4 ia Glycogen Synthow (0) Phaphoryloaa 027 Ser-Arq-Pro-La"-Ser-Asp-Wn-Flu-L a Arq-L a Gln-Ila -Ser-Vol-Ar GI Leu I2 i 2 3 4 5 a r a io 11 12 is 14 is ie Ssr-Asp-GIn-GIu-Lyr-A~~-Lye-GIn-Ils- Ser-Val-Arq-GI Leu I2 5 a 7 a 9 ii 12 13 14 is ie J - i s

11- 11-

I

I

I

09 09 17 09

0.18 225

2

0 32

3 4

086

5

(a)

I50 39

6

29

7

29 27 055

8

9 10

058

II

10

018 012

12 13

08 09

052 025

14

09

17 049

17 18 19 20

23

15 16

IS

09 02

088

02 02 08

038 0029 0019

21 22

148

184

156

323 237

309 007 041 088 2 33 006 0 I4

25 26

0.70

0.04

001

23 24

+

II C I

021 030 218 057

27

28

29 30 31 32 33

M

FIG. 2. Peptide substrates for phosphorylase kinase. Data taken from Refs. (81-84). Values of

K,,,and v,

denoted by indicate that the activity with this substrate was too low to be measured. (a) In separate studies (Sa, the K,,,for glycogen synthase has been reported to be approximately the same as for phosphorylasebut the V,,, is about one-half. (b) The sequence of peptides derived from glycogen synthase are aligned with those from phosphorylase with residues 6-10 deleted. (c) The data for peptides 25-34 were from a study separate from that for peptides 7-24. There were minor differences in the values from the two studies, as indicated by data for the peptide designated 18 in the first study and 25 in the second. (d) Data from (85). "-'I

The primary conclusions that can be derived from these studies are as follows:

1. In the phosphorylase sequence, little change occurs when the first eight Nterminal amino acids are deleted from the native sequence (peptide 8); however, deletion of the two carboxyl-terminal amino acids (Gly- 17Leu-18) dramatically reduces the rate of phosphorylation, albeit that the K,,,values are more analogous to that of phosphorylase (peptides 19-22). 2. The simple substitution of the phosphorylatable serine by threonine markedly reduces phosphorylation of the peptide despite the presence of the needed hydroxyl group (peptides 24 and 34).

415


3. There appears to be a requirement for basic amino acids to be present on both (or either) the N- and C-terminal sides of the phosphorylatable serine, although the results are ambiguous. Thus, replacement of either Arg-10 or Lys-11 or both (peptides 9-1 1) in the phosphorylase sequence only modestly affects activity, whereas replacement of Arg-4 in glycogen synthase by Lys (peptide 2) makes it a poor substrate, and by Leu (peptide 1) eliminates activity. Similarly, for the potential requirement of a basic amino acid on the C-terminal side, the native glycogen synthase does not contain a basic amino acid in that position, but replacement of Ser-9 by Arg (peptide 3) markedly improves the peptide as a substrate and substitution of Arg-16 in the phosphorylase sequence by Ala (peptide 12) or Gly (peptide 13) markedly diminishes the V,,,,,. Another residue of potential importance in dictating substrate specificity appears to be Gln-12 in phosphorylase, since Asn replacement (peptide 27) markedly decreases the v,,,,,; however, in glycogen synthase the equivalent position contains a Thr. A hydrophobic residue on and directly next to the C-terminal side of the phosphorylatable serine also appears essential; in both glycogen synthase and phosphorylase this is Val; substitution by Ile (peptide 15) markedly reduces the rate of phosphorylation. These studies have thus begun to provide information on what dictates substrate specificity of phosphorylase kinase, but in all probability more is involved than the sequence of amino acids. A full elucidation of substrate requirements will most likely require a study, not only of the sequences of the peptides but also of their conformational structure and what conformation they can assume when associated with the enzyme. That phosphorylase is a better substrate than any of the peptides clearly indicates that tertiary structure plays a role in establishing the efficacy of a substrate. In addition to phosphorylase b and glycogen synthase (86, 89, 90, ~ O U ) , phosphorylase kinase can, in v i m , phosphorylate itself (18, 61, 62, 87, 88), and has been reported to phosphorylate troponin I (91), troponin T (92, 93), the sarcolemmal Na ,K ATPase ( 9 4 , the Ca2 -dependent (transport) ATPase of sarcoplasmic reticulum (32, 33,) casein (94a), myosin light chain (94b), and several other proteins. It is of interest that the site phosphorylated on glycogen synthase (95) and on the P-subunit in autophosphorylation (88) are also phosphorylated by the CAMP-dependent protein kinase, whereas the latter enzyme does not phosphorylate phosphorylase. As with glycogen synthase (see Fig. 2), the site phosphorylated on the P-subunit of phosphorylase kinase does not contain an Arg residue on the C-terminal side of the phosphorylated serine. AIthough peptides containing Thr instead of Ser are not readily phosphorylated, the site phosphorylated in troponin I is threonine. Whether, in addition to phosphorylase, any of these proteins, or others, are indeed physiological substrates for phosphorylase kinase is not known. The site +

+

+

416

CHERYL A. PICKE’IT-GIES AND DONAL WALSH

phosphorylated on glycogen synthase in virro can be phosphorylated within the cell (96), but since this site is also phosphorylated by at least three other enzymes, it is not known which one(s) are responsible for the phosphorylation in vivo.It will be difficult to prove whether or not glycogen synthase is indeed a cellular substrate for phosphorylase kinase. Currently, it appears unlikely that phosphorylase kinase autophosphorylation occurs physiologically. Were it to occur, it would most probably do so in response to an elevation of Ca2+ since autophosphorylation is Ca2 -dependent. However under in vivo conditions, where Ca2+-dependent phosphorylase kinase-catalyzed phosphorylation of phosphorylase clearly occurs (73),there is no indication of phosphorylase kinase autophosphorylation (and autoactivation). This has been explicitly examined for both subunits in perfused cardiac muscle (96a). Two attributes of phosphorylase kinase suggest that it may well have functions in addition to the regulation of phosphorylase. First, as previously indicated, phosphorylase kinase has a very complex structure and, in comparison to other enzymes, it appears to be more complex than is necessary simply for the regulation of glycogenolysis. Second, phosphorylase kinase is present in skeletal muscle at a very high concentration (0.5%of soluble protein). Since phosphorylase is 2% of the soluble protein, then, using the molar activities for the nonactivated enzyme measured by Chan and Graves (45) (99.3 molecules/sec), it can be calculated that the amount of phosphorylase kinase present in skeletal muscle would be sufficient to fully activate all of the phosphorylase in the cell within one-tenth of a second. Activation (by covalent modification) would presumably increase this rate even more. Thus it appears that the amount of phosphorylase kinase present in muscle is far greater than is needed to activate glycogenolysis, even under the most extreme circumstances. It is of interest that whereas phosphorylase, when isolated, is totally associated with the glycogen particle, a different result is observed with phosphorylase kinase, where 20-30% is glycogen bound and most of the rest is cytosolic (97). Jennissen et al. (98) have shown by cytochemical techniques that a “phosphorylase kinase” has a localization distinct from that of phosphorylase, with most phosphorylase appearing to be glycogen bound but most phosphorylase kinase being present in the region of the sarcolemma. Dombradi et al. (99) have further shown that purified rabbit muscle T-tubules contain phosphorylase kinase, and Le Peuch et al. (33) and Horl and Heilmeyer (32) reported that a unique form of “phosphorylase kinase” is associated with purified sarcoplasmic reticulum vesicles. All of these data suggest that phosphorylase kinase may well play some role in addition to the regulation of phosphorylase and, in particular, that the sarcolemma Na ,K -ATPase and the sarcoplasmic reticulum Ca2 dependent ATPase may be target sites for control. If so, then phosphorylase kinase may have an important role in ionic homeostasis as well as metabolite availability. +

+

+

+

417


SUBSTRATE SPECIFICITY D. NUCLEOTIDE The nucleotide specificity of phosphorylase kinase has been examined by Flockhart et al. (100) by comparing it with the CAMP-dependent and cGMPdependent protein kinases. As might be expected from the sequence homologies of the catalytic sites (28),the three enzymes are similar with respect to nucleotide affinities. Comparisons between nucleotides were made by competition with ATP. The principal conclusion derived is that the 6-amino and P-phosphoryl groups are primary factors in dictating substrate specificity. No other natural nucleotide triphosphates serve as suitable substrates.

V. Regulation of Phosphorylase Kinase Activity by Allosteric Effectors Phosphorylase kinase activity can be modulated by a number of effectors that interact in a noncovalent and specific manner and presumably modulate activity by affecting the enzyme's conformation. Included in this group are Ca2+, Mg2+, calmodulin, and glycogen; each is discussed here. Phosphorylase kinase activity is also affected by pH, ADP (see Section IV,A), actin (100a, see also Section V,C), ionic strength, several phosphate-containing compounds, and organic solvents. The latter have been discussed in detail elsewhere (11) and need little additional comment. One comment is pertinent, though, particularly in the area of ions used and ionic strength. The conditions that various investigators have used to investigate phosphorylase kinase have varied widely. Ionic strength, and the type of ions present, have a marked effect on phosphorylase kinase structure, and which of these conditions truly mimics the conformation in which phosphorylase kinase exists in the cell is not known; inappropriate conditions can clearly lead to artifactual observations. Personal bias suggests that phosphorylase kinase is particularly sensitive to such changes. In many of the past studies glycerophosphate has been employed as buffer. In glycerophosphate, however, enzyme activity, the degree of phosphorylation, and the regulation by ADP are each suppressed in comparison to some other buffer system; thus glycerophosphate is clearly not the buffer of choice for several types of studies. A.

CA2'

The requirement of phosphorylase kinase for Ca2 was first demonstrated by Meyer el al. (70). EGTA, a relatively specific chelator, potently inhibited the enzyme's activity, and this inhibition could be reversed by the addition of excess Ca2+ ions (7,70). Besides being required for the phosphorylation of phos+

418

CHERYL A. PICKETT-GIESAND DONAL WALSH

phorylase b, Ca2 has since been shown to be necessary for the phosphorylation of glycogen synthase, troponin, and phosphorylase kinase itself (61, 62, 87, 88, +

101).

The mechanism by which Ca2+ regulates phosphorylase kinase is clearly allosteric. Not only is the 6 subunit identified as a Ca2+-binding subunit whose properties can fully account for Ca2+ binding by the holoenzyme (see Table I), but several lines of evidence have shown that Ca2 does not participate directly in catalysis. Included in the evidence are the observations that (a) free y-subunit, obtained by partial dissociation and EGTA treatment plus sucrose gradient centrifugation (46),exhibits catalytic activity that is independent of Ca2+, and (b)in most studies some EGTA-insensitive activity exists (54, 74). Allosteric activation of phosphorylase kinase by Ca2 promotes as much as a 25-fold change in the K,,,for phosphorylase (102). Removal of Ca2+ by EGTA addition prompts dissociation of the phosphorylase kinase-phosphorylase complex (103). The activation of phosphorylase kinase by Ca2 requires the binding of Ca2 to at least three of the four Ca2+ binding sites per 6-subunit (i.e., at least 12 mol per holoenzyme) (104, 105). Thus, for nonactivated phosphorylase kinase (i.e., dephospho-enzyme), allosteric activation (in the presence of Mg2 ) requires Ca2+ in the concentration range of 2-25 pA4 (54, 104, 105). [Some of the variation between reports most probably reflects differences in experimental conditions; another likely cause is varying minor degrees of proteolysis (106)]. This level of Ca2+ which is required to activate phosphorylase kinase coincides well with the binding constants determined by direct binding studies (Table I), given the differing conditions needed to study the two parameters. The K, for Ca2+ for phosphorylase kinase activation is reduced about 15- to 30-fold by phosphorylation of the enzyme by the CAMP-dependent protein kinase (106, 107-109) and as much as 300-fold by proteolysis (106). [Here again, there is some discrepancy between the amount of change reported; see Ref. (54) for example.] With cardiac phosphorylase kinase (a' isozyme), in contrast to the skeletalmuscle enzyme, phosphorylation does not appear to modify the requirement for Ca2+ (22, 72). The activation of phosphorylase kinase by Ca2 clearly occurs physiologically. In resting skeletal muscle, the intracellular concentration of Ca2+ is in the range of 10-100 nM, which rises to 1-10 pkf upon stimulation of muscle contraction (103, 104). These concentration changes are in the appropriate range to allosterically regulate phosphorylase kinase and thus provide a physiological link whereby contractile activity is connected to enhanced glycogenolysis. [Evidence that this indeed occurs in skeletal muscle has been well documented, initially by the work of Drummond et al. (110) and Stull and Mayer ( I l l ) ,and subsequently verified by several investigators.] As discussed in more detail in Section VIII,A, in electrically stimulated muscle, enhanced contraction is clearly associated with the phosphorylation and activation of phosphorylase, without +

+

+

+

+

+

419


covalent modification of phosphorylase kinase, and without an increase in CAMP or CAMP-dependent protein kinase activity. The cause of this phosphorylase activation appears to be the stimulation of phosphorylase kinase by increased cytosolic Ca2 . Activation of phosphorylase by the Ca2 -dependent stimulation of phosphorylase kinase has also been well documented as the mechanism underlying the a-adrenergic activation of glycogenolysis in both heart (73) and liver (112-115). In skeletal muscle, the primary organelle that sequesters Ca2+ is the sarcoplasmic reticulum. Addition of isolated sarcoplasmic reticulum to phosphorylase kinase in v i m inhibits its activity, and this inhibition can be reversed by the addition of Ca2+ (107). More details on the regulation of phosphorylase kinase by Ca2+ are presented in Sections V,C and VII,A. +

B. Mc2

+

AND

+

Mc2 - C A ~ INTERACTIONS +

+

One major role for Mg2+ in phosphorylase kinase-catalyzed reactions is to serve as part of the substrate, MgATP2-. In addition to this, however, Mg2+ modulates phosphorylase kinase activity in a manner that is often manifested as an interaction between Mg2+ and Ca2+. These interactions have been studied but have been difficult to unravel. A likely site of Mg2+-binding to phosphorylase kinase is the 8-subunit. Isolated calmodulin binds Mg2 at the Ca2 binding sites and, although the binding of Mg2 to calmodulin is several orders of magnitude weaker than that of Ca2 , the two interact with calmodulin at physiological concentrations that are the same relative concentrations that affect phosphorylase kinase. Interactions between Mg2 and Ca2 on the &subunit could readily explain the variety of effects that have been observed for these two ions on phosphorylase kinase. Whether, in addition to those on the &subunit, there are other specific Mg2+-binding sites on phosphorylase kinase is not known. One particular problem in interpreting the results obtained so far is that different investigators have most frequently used different assay conditions. This has made comparisons difficult since ionic strength, pH, and types of ions all seem to affect the responses observed with varying concentrations of Mg2 and Ca2+. In one of the most complete studies, Kilimann and Heilmeyer (54) described the effects of Ca2+ and Mg2+ in terms of three types of activities, designated &, A,, and A,, according to their dependence on Ca2 , Mg2 , ionic strength, and pH. These do not necessarily represent unique catalytic sites, although that has been suggested as one possible explanation. At a minimum, these three activities reflect different conformations of the enzyme brought about by the presence of the various ions and cations. In the Kilimann and Heilmeyer terminology, the A, activity is only a small portion (0.1-1%) of the total activity of nonactivated phosphorylase kinase and is essentially.Ca2+-independent. Most likely, this activity is identical to that +

+

+

+

+

+

+

+

+

420


described by King and Carlson (74, 75) as being “EGTA-insensitive.” As discussed in Section IV,B, King and Carlson (74) report that the lag that is characteristic of the phosphorylase kinase-catalyzed phosphorylase conversion progress curve can be removed by preincubation of the enzyme with Mg2 plus Ca2+. This preincubation results in a 2- to 7-fold increase in total (Ca2+dependent) activity [depending upon conditions (74, 75)], and can account for the removal of the lag in the progress curve. In addition to this effect on total activity, however, this preincubation with Ca2+ and Mg2+ also results in up to a 30-fold increase in the EGTA-insensitive (A,) activity (75). Several characteristics described by King and Carlson (75) suggest that the activation of the total (Ca2 -dependent) activity and of the EGTA-insensitive activity are not fully the same process. In addition to the A, activity, Kilimann and Heilmeyer (54) described two other activities, A, and A,, that are Ca2+-dependent. The A, activity constitutes most of that seen at pH 6.8 and is a high-affinity Ca2+-dependent activity (KO, Ca2+ = 1.4 pM), which requires free Mg2+ ( K , = 0.25 mM at pH 6.8). High Mg2+ inhibits this activity (Ki = 3.5 mM) due to competition with Ca2+ ions. The A, activity is a low-affinity Ca2+-dependent activity that at pH 6.8 is induced by 20 mM Mg2 and is more prominent at high pH due to an increase in V,,, and an increased affinity for Mg2 . The A, activity is half-maximally stimulated by Ca2+ concentrations of 10-70 pM (depending on Mg2+ concentration and pH). All three activities (A,, A,, and A,) are inhibited at millimolar concentrations of Ca2+ and this inhibition is competitively antagonized by Mg2 . Kilimann and Heilmeyer (54,55) have also presented evidence suggesting that the three types of phosphorylase kinase activities have different functions. All three catalyze the phosphorylation of phosphorylase but the characteristics of cation dependencies suggest that autophosphorylation is only catalyzed by A, and A,, troponin I phosphorylation only by A, and troponin T phosphorylation only by A,. The three activities also appear to be differentially regulated. Cyclic AMP-dependent phosphorylation (of only the p subunit or of both a and p subunits) appears to regulate only the A, activity, autophosphorylation (of a plus p subunits) appears to stimulate both the A, and A, activities, and limited trypsin proteolysis increases the V,,, of all three and the Ca2 affinity of A,. The exact nature of the A,, A , , and A, activities of phosphorylase kinase remains to be resolved. Perhaps they reflect different catalytic sites, but it is equally possible that they represent three different conformations of the protein brought about by the relative concentrations of different cations and manifested as different interactions at a single catalytic site. In a further report from this same laboratory (55b),the A,, A , and A, activities have been proved using subunit-specific monoclonal antibodies. One anti-a antibody obtained provoked a marked enhancement of Ca2+independent activity (A,) without affecting Ca2 -dependent activity; it was suggested that the antibody might in some way uncouple an inhibitory signal +

+

+

+

+

+

,

+

42 I


between Ca2+-free 6 subunit(s) and the catalytic site(s). Analogously, an anti-6 subunit antibody was shown to block Ca2+-dependent activity suggesting that restriction on 6 subunit conformation can block 6 subunit-catalytic site signal transfer. These studies would appear to be a beginning probe in studies of the major conformational changes and interactions which are clearly integral to the multiple interactions that exist for the regulation of this enzyme. Other investigations of the effects of Mg2+ on phosphorylase kinase (65-69) have provided conclusions that generally concur with those presented by Kilimann and Heilmeyer (54, 55). Singh el al. (67) studied the effects of varying concentrations of Mg2 , either in the presence of maximal Ca2 or with Ca2 absent. With Mg2+ in substantial excess of ATP, varying Mg2+ resulted in a 5to 10-fold activation of nonactivated phosphorylase kinase (measured at either pH 6.8 or 8.2, and at pH 6.8 in the presence or absence of Ca2+) and of phosphorylase kinase activated by either the CAMP-dependent protein kinase, autophosphorylation, or limited trypsin proteolysis. This stimulation is accounted for by a decrease in the K , for both phosphorylase and ATP and an increase in V,,,, and is apparently due to a direct interaction with phosphorylase kinase since it is also observed with casein as substrate. Hallenbeck and Walsh (28) compared the effects of Mg2+ and Mn2+ on phosphorylase kinase-catalyzed phosphorylase phosphorylation and autophosphorylation. Varying Mg2 affects both in the same concentration range, whereas varying Mn2+ stimulates autophosphorylation but inhibits phosphorylase phosphorylation. This appears to provide a distinction in the role that the metal ions play in these two processes that cannot be attributed simply to allosteric effects modulating a single catalytic site. As discussed in Section IV,B, King and Carlson (74, 75) have shown that the synergistic presence of both Ca2+ and Mg2+ causes a slow conformational change in the structure of phosphorylase kinase that leads to its activation. Using cross-linking agents as a probe, the conformational change induced has been shown to be similar to that which occurs upon activation by phosphorylation, proteolysis, and high pH (38). The full relationship between this synergistic effect of the two divalent cations, the three states of phosphorylase kinase termed by Heilmeyer &, A,, A,, the binding characteristics of Ca2+ and Mg2+ (Table I), and the effects of phosphorylation and proteolysis on divalent cation binding clearly needs considerable clarification. Although it is clear that varying Mg2 affects phosphorylase kinase activity, the physiological ramifications of this observation are not apparent. Free Mg2 in muscle is approximately 4 mM (116) and is thought to remain relatively constant. It would certainly be surprising if it varied to an extent that would cause significant differences in phosphorylase kinase activity. The role of Mg2 thus appears to be more that of a constitutive cofactor that is either bound or not, depending upon other factors. Heilmeyer et al. (227), for example, proposed a +

+

+

+

+

+

+

422


mechanism wherein CAMP-dependent phosphorylation of phosphorylase kinase influences Mg2 -binding properties of the enzyme, which in turn modifies Ca2 binding and hence Ca2 -dependent regulation. In other words, in the unactivated state, the Ca2 -binding sites on the 6 subunit may, as their “counter-ion,’’ be occupied by Mg2+ that binds but does not produce an active conformation. Increases in Ca2 promote exchange, or exchange might arise because the relative affinities for Ca2 and Mg2 are modified by other changes occurring on the protein, such as covalent modification. The effects of Ca2+ and Mg2+ on phosphorylase kinase could all be explained by such a mechanism given the known cooperativity between binding sites (either on the same 6 subunit or different subunits). Thus Mg2+ may be an important constituent of phosphorylase kinase in the cell and be involved in its regulation, even though it does not itself change in concentration. +

+

+

+

+

+

+

BY EXTRINSIC CALMODULIN AND TROP~NIN C C. REGULATION

As previously described, phosphorylase kinase activity is regulated by Ca2 , mediated via the 6 subunit whose structure is identical to calmodulin except for the presence of trimethyllysine and carboxylation. In addition to this intrinsic calmoddin, however, skeletal muscle phosphorylase kinase can be specifically activated by extrinsic calmodulin (15, 37, 89, 106, 118, 119). This activation also requires Ca2+ but, unlike the intrinsic Ca2 -dependent activation occurring via the 6 subunit, activation by extrinsic calmodulin is blocked by addition of other calmodulin-binding proteins (which compete for calmodulin) and by phenothiazines. In addition, when Ca2+ is removed, the extrinsic calmodulin does not remain bound to the holoenzyme. Cohen (106)has termed this extrinsic calmodulin the “6’-subunit.” Over and above the almost total requirement of holophosphorylase kinase for Ca2 , addition of extrinsic calmodulin activates the enzyme, synergistically, a further 2- to 7-fold. Extrinsic calmodulin has no effect on phosphorylase kinase in the absence of Ca2 . Activation by extrinsic calmodulin is observed not only with phosphorylase b as substrate, but also in the phosphorylation of glycogen synthase (89),phosphorylase kinase itself (12, 89), and troponin I (12). The activation by exogenous calmodulin is very pH dependent. The activation of phosphorylase kinase by extrinsic calmodulin occurs via a mechanism very similar to that observed for other calmodulin-dependent enzymes, albeit with the added complexity that the enzyme itself, in the absence of extrinsic calmodulin, also binds and is activated by Ca2+. A scheme for the activation of phosphorylase kinase by Ca2+ and extrinsic calmodulin, taken in particular from the works of Burger et af. (104) and Cohen (106), is presented in Fig. 3. In the absence of extrinsic calmodulin, phosphorylase kinase activation requires the binding of at least 3 mol of Ca2+ per intrinsic 6-subunit. In the +

+

+

+

423


( a P r S - C o ~ + ) lntrinsic 3 4

Activation

(aPyS-Cay)4 (CoM-Coy)4 lntrinsic

a Extrinsic Activation

FIG. 3. Scheme for activation of phosphorylase kinase by calcium and calmodulin. Activation by Ca2+ with either intrinsic or extrinsic calmodulin requires a minimum of three of the four Ca2+binding sites to be occupied. Modified from the data of Burger et al. (104) and Cohen (106).

presence of Ca2 , phosphorylase kinase also binds extrinsic calmodulin stoichiometrically [i.e., one mole of calmodulin per a@$ (37) with a Kd of 2-15 nM (37, 89, 104, 106)].Similar to observations with other calmodulin-requiring enzymes, the binding of extrinsic calmodulin occurs in two steps; namely, initial binding of at least 3 mol of Ca2+ per mol of calmodulin, followed by binding of the (Ca2 ),-calmodulin complex to the enzyme. The binding of extrinsic calmodulin to phosphorylase kinase does not appear to affect the binding characteristics of Ca2+ to the intrinsic &subunit, and the binding characteristics of Ca2 to either the intrinsic &subunit or extrinsic calmodulin appear to be identical. As a consequence, addition of extrinsic calmodulin does not alter the Ca2+ concentration dependence of phosphorylase kinase activity but only the maximum velocity; that is, (apy8-Ca2 3)4 and (ol@y8-Ca2 3)4 (CaM-Ca2 3)4 have the same Ca2+ concentration dependence for formation but the V,,, of the calmodulin activated form, is 2- to 7-fold greater than V,, of the simple holoenzyme. Extrinsic calmodulin has no effect on the affinity of phosphorylase kinase for ATP, phosphorylase, or glycogen (89). Burger et al. (104) suggested that the inactivation of phosphorylase kinase activated by extrinsic calmodulin does not follow a simple reversal of the activation process. The most likely route of inactivation appears to be first the dissociation of Ca2+, followed by the dissociation of extrinsic calmodulin. The extrinsic calmodulin and the intrinsic 6 subunit, however, maintain their separate integrities, and no exchange occurs (within the typical activation-inactivation process) between the two species. The activation of phosphorylase kinase by extrinsic Ca2+-binding protein occurs not only with calmodulin but also with the homologous protein, skeletal muscle troponin C (104). Interestingly, cardiac troponin C, which, though also homologous exhibits distinct Ca2 -binding characteristics (120-Z22), does not serve in this role. This indicates that the activation by the skeletal muscle tro+

+

+

+

+

+

+

424


ponin C has a high degree of specificity. Activation is also not observed with the parvalbumins, though they too are homologous with calmodulin (106). The characteristics of activation of phosphorylase kinase by troponin C are similar to those with calmodulin but with some important differences. Activation by troponin C requires much higher concentrations (Kd = 1-2 pM) than that by calmodulin (Kd = 2-15 nM), but troponin C activates to a greater extent (20- to about 30-fold) and, most importantly, appears to increase the sensitivity to activation by Ca2+ by about 5-fold (106). Cohen (106) has argued that troponin C, rather than calmodulin, may well be more physiologically important as the extrinsic activator of phosphorylase kinase. One reason for this proposal is that, while a 100-fold higher concentration of troponin C is required for activation, it is present in the cell at a concentration (100 pJ4) greater than that required for phosphorylase kinase activation. Potentially, it may be more readily available than the more limited supplies of calmodulin. More importantly, the concentration of Ca2+ with which troponin C activates phosphorylase kinase appears to be in a more physiological range than that required of extrinsic calmodulin. Two additional facts support Cohen’s proposal. First, he has shown that artificial thin filaments, composed of actin, tropomyosin, and the troponin subunits, are equally as effective as isolated troponin C in the activation of phosphorylase kinase. This observation is particularly pertinent since most (if not all) troponin C in the cell is bound as a component of the myofibrils. Second, Sigel and Pette (123, 124) have shown that the protein-glycogen particle (of which phosphorylase kinase is one component) appears to be associated with the myofibrils. This suggests that the localization of both phosphorylase kinase and troponin C within the cell is highly suitable for one to be regulating the other. An interesting adduction to this is the observation of Livanova et al. (1OOa), who report that phosphorylase kinase is also separately activated by actin, shown to be through promotion of an increased affinity for phosphorylase and an increased V,,,,,. It appears to be the polymerized F-actin form which is the effective activator. The activation of phosphorylase kinase by extrinsic calmodulin is isozymespecific. This has been shown by Tam et al. (21) for the a’ isozyme isolated from bovine red-skeletal muscle, and by Yoshikawa et al. (125)for this isozyme isolated from rabbit heart, neither of which is activated by addition of calmodulin. Some activation of the heart enzyme has been seen with skeletal muscle troponin C, but since no activation is observed with cardiac troponin C, this suggests that the a’isozyme is not regulated physiologically by extrinsic Ca2 binding proteins. The a’ isozyme is regulated by Ca2 in a manner identical to the a isozyme via the intrinsic &subunit (21, 22). Yoshikawa et al. (125) suggested that this difference in regulation between the two isozymes by extrinsic Ca2 -binding proteins may reflect a difference in the needs of the different muscle types for a tighter regulatory linkage between contraction and glycogenolysis. Thus, in fast-twitch muscle, which contains the a isozyme of phos+

+

+

425


phorylase kinase, contraction relies heavily on glycogenolysis as an energy source. In contrast, in muscles with higher oxidative capacity (cardiac, and FOG and SO skeletal muscle), which contain the a‘ isozyme, glycogen metabolism is much less important as a source of energy because of a higher aerobic metabolic potential. The difference in the binding of calmodulin to the a and a‘ isozymes currently serves as the best experimental procedure to separate the two isozymes during purification (15, 102). Of potential note, the skeletal muscle a isozyme, when activated by phosphorylation by the CAMP-dependent protein kinase, is also not significantly activated by extrinsic calmodulin or troponin C but displays an increased sensitivity to Ca2 activation compared to the nonphosphorylated enzyme (106).Thus, epinephrine stimulation enhances glycogenolysis directly, without the need for additional activation by extrinsic Ca2+-binding proteins. The binding of extrinsic calmodulin to holophosphorylase kinase is quite distinct from that of the intrinsic &subunit. Cross-linking studies (37) suggest that its most likely sites of binding are on the a-and &subunits (see Fig. 1) and this conclusion is supported by studies of the effects of proteolysis. Apparently the a-subunits must play a distinct role in the interaction with extrinsic calmodulin since no binding occurs with the a’isozyme. Phosphorylase kinase displays an essentially negligible rate of exchange between the intrinsic and extrinsic calmodulin with a t,,, in the neighborhood of several weeks (37). +

D. REGULATIONBY GLYCOGEN AND IN THE PRESENCE OF THE CONSTITUENTS OF “GLYCOGEN PARTICLE” Within the cell it appears likely that most, if not all, of the phosphorylase kinase-catalyzed activation of phosphorylase occurs within a complex containing these two enzymes, and others, associated with glycogen. By electron microscopy, glycogen in the cell has been identified as being present in isodiametric particles, termed P-particles, of 150 to 300 A in diameter, having unique subcellular localization. This is the typical form of glycogen in muscle where glycogen is associated with the sarcoplasmic reticulum and myofibrils. In liver, notably, these P particles form aggregates of a larger size, termed a particles, that are rosette-like in appearance. Such glycogen particles have been isolated from muscle by a variety of methods relying primarily on procedures of differential centrifugation (97, 102, 126-128). The isolated glycogen particles contain all of the enzymes of glycogen metabolism plus sarcoplasmic reticulum vesicles. The high degree of consistency in composition of the glycogen particles prepared by different procedures strongly suggest that they do indeed represent the organization of the enzymes of glycogen metabolism as occurs within the cell. This conclusion is also supported by the observation that the morphology of the isolated glycogen particles is very similar to the structures observed in the intact cell. Wanson and Drochmans (126) have found evidence for specific structures

426

CHERYL A. PICKE'lT-GIES AND DONAL WALSH

that link the particles to the sarcoplasmic reticulum. The enzymes of glycogen metabolism can be dissociated from the sarcoplasmic reticulum membranes by amylase digestion of the glycogen (97);however, a glycogen particle containing these enzymes but dissociated from sarcoplasmic reticulum vesicles has not been obtained. Using direct precipitation using either ultracentrifugation or acetone typically, glycogen particles contain 70-90% of the glycogen synthase,' 70100% of the phosphorylase, and 20-40% of the phosphorylase kinase found in skeletal muscle. When an acid precipitation step is also included a lower amount of glycogen is associated with a lower ratio of phosphorylase to phosphorylase kinase, suggesting that there are possibly different species of glycogen particles. Both CAMP-dependent protein kinase and protein phosphatases appear to be specifically associated with glycogen particles, but, fully consistent with the roles they play in other regulatory events within the cell, much of the activity of these enzymes is present in other subcellular locations. It may be of particular significance that only a fraction (20-40%) of phosphorylase kinase is associated with these particles (97);this observation suggests that phosphorylase kinase has other functions within the cell in addition to phosphorylase activation. Depletion of glycogen in vivo results in the release of the enzymes of glycogen metabolism from the glycogen particle to the cytosol (127). Of note, even following extensive characterization ( 1 2 8 ~there ) appears to be some residual catecholaminesensitive adenyl cyclase associated with the SR-glycogen particle. It appears clear that the phosphorylase kinase-catalyzed activation of phosphorylase occurs in the cell with part or all of these enzymes not being freely soluble in the cytosol, but within a complex with a specific organization of the component parts. Because of this, it becomes particularly pertinent for the reactions involving phosphorylase kinase to be examined with as much of this organization intact as possible. Phosphorylase specifically binds glycogen and studies of its 3-dimensional structure have revealed that, in addition to the substrate site, there is a unique glycogen-binding site that is not involved in the catalytic process (129, 130). Most likely, this is one of the specific interactions that forms the basis of the organization within the glycogen particle. In addition to this, phosphorylase kinase also binds glycogen. Initially, this was recognized from the studies of DeLange et al. (8) who observed a Mg2 -dependent glycogen-phosphorylase kinase complex, and also found that phosphorylase kinase autophosphorylation was stimulated by glycogen addition (8). Steiner and Marshall (131)have shown that there is a synergistic effect of both Ca2 and Mg2 upon the interaction of phosphorylase kinase with glycogen, and we have subsequently found that this is also true for the formation of a glycogen phosphorylasephosphorylase kinase complex (132). Whether this divalent cation synergism is similar to that reported by King and Carlson (53, 74) has not been evaluated. The effects of glycogen on phosphorylase kinase have been examined by two approaches, one employing purified enzymes, the other the isolated glycogen +

+

+

427


particle. From both, however, our current knowledge is still only fragmentary. With the isolated enzyme, glycogen activates phosphorylase conversion 3- to 6fold (8,22); maximum stimulation occurs with about 0.3% glycogen. Activation is greater at pH 6.8 than at pH 8.2, and arises as a consequence of up to a 12-fold decrease in the K,,,for phosphorylase b. Since Tabatabai and Graves (69) found that glycogen has no effect on phosphorylase kinase-catalyzed peptide phosphorylation, it is most likely that glycogen acts via interaction with the substrate, phosphorylase, rather than the catalyst, phosphorylase kinase. Studies of phosphorylase activation in the glycogen particle were first initiated by Fischer’s laboratory (97, 102). As previously discussed, the glycogen particle contains not only the enzymes of glycogen metabolism, but also sarcoplasmic reticulum vesicles. The latter have a maintained and highly active Ca2+ accumulation capacity which leads to regulation of Ca2+ concentration in the media in which the glycogen particles are suspended. Fischer’s laboratory studied what is termed the “flash activation” of phosphorylase in these particles. In the isolated glycogen particle, even if incubated at 30°C with MgATP2-, phosphorylase and phosphorylase kinase are inactive and the rate of glycogen degradation minimal. The total lack of activity of phosphorylase kinase reflects the complete removal of any free contaminating Ca2 by the sarcovesicular system. This situation is probably an accurate reflection of what occurs in unstimulated skeletal muscle, where the sarcoplasmic reticulum system dominates the cytosolic Ca2 concentration. Incubation of the glycogen particle with Ca2-I--EGTA buffers plus MgATP2- results in flash activation, characterized by a rapid activation of phosphorylase b to a, and a subsequent rapid reversal to basal levels. Typically, with a glycogen particle containing 40 mg/ml protein, the addition of 10 kmol of ATP results in the conversion of 50-80% of phosphorylase b to a within 15 sec, and a return to basal levels within 3 min. The duration of phosphorylase activation is increased by increasing ATP, but under all circumstances inactivation is concomitant with the depletion of ATP, the latter being caused primarily by the Ca2 -dependent ATPase of the sarcoplasmic reticulum Ca2+ transport system. All of the properties of flash activation are consonant with the activation being due to allosteric activation of phosphorylase kinase by Ca2 . The Ca2 requirement for this activation appears to be somewhat higher in the glycogen particle than with the isolated enzyme (102). Since a-amylase digestion results in the Ca2 -dependency becoming similar to that for the isolated enzyme, it suggests that glycogen, or the organizational structure of the glycogen particle, modifies the divalent cation dependency of the enzyme +

+

+

+

+

+

(102).

Most of the additional studies on the properties of the glycogen particle report either on the regulation of phosphorylase kinase phosphorylation, which is described in subsequent sections of this review, or on phosphorylase, which is beyond the scope of this review but has been reviewed elsewhere (133, 134).

428


However, three further observations of phosphorylase kinase activity in the isolated glycogen particle are of potential interest: (a) Srivastava et al. (135) reported a Ca2 -dependent inactivation of glycogen synthase in these particles that, on the basis of antibody studies, could be attributed to phosphorylase kinase-catalyzed phosphorylation. (b) although in their initial report Heilmeyer et al. (102) did not observe Ca2+-dependent autophosphorylation of phosphorylase kinase, we have demonstrated that this can occur (132). It can be masked, however, by concomitant Ca2 -Mg2 -dependent synergistic activation. (c) Heilmeyer e?ad. (102)reported that, in the glycogen particle, as initially described by Fischer and Krebs (137) with the isolated enzyme, conversion of phosphorylase b to a involves the stepwise addition of phosphate with the formation of intermediary phospho-dephospho hybrids. The potential relevance of these hybrids in metabolic control has, for the most part, been ignored (136). From the fragmentary studies that have been performed with glycogen particles isolated from skeletal muscle or cardiac muscle (138, 139), it seems clear that the organizational structure imposed upon phosphorylase kinase by the glycogen particle is of fundamental importance in dictating the reactions that occur. The ramifications of such controls have yet to be fully realized. +

+

+

VI. Proteolytic Activation of Phosphorylase Kinase The activity of phosphorylase kinase towards phosphorylase b can be stimulated by at least two types of covalent modification, phosphorylation and proteolysis. Early observations of Ca2 -dependent activation of partially purified phosphorylase kinase were found to be due in large part to the presence of a contaminating Ca2 -dependent protease, referred to in the early literature as “kinase-activating factor” (KAF)and later as “calcium-activating factor” (CAF) (70).The proteolytic activation was initially distinguished from that due to direct Ca2+-allosteric action on the kinase in that the former required millimolar concentrations of Ca2+, was time dependent, and was irreversible. The Ca2 -concentration dependency is no longer a suitable means for distinguishing the two mechanisms since proteases have since been described that are activated by Ca2+ in the more physiological, micromolar range (140). Other proteases, such as trypsin, chymotrypsin, and papain are also capable of activating phosphorylase kinase. The high sensitivity to proteolysis, which can readily be demonstrated at very low protease : protein concentration ratios, in contrast to the lack of specificity as to the type of peptide bond cleaved, suggests that phosphorylase kinase must contain highly exposed regions of peptide chain with little conformational organization, yet which must play a major role in regulating activity. Increases in activity due to proteolysis are on the order of 50- to 100fold at pH 6.8 and 2- to 3-fold at pH 8.2 (11). The a subunit is the most +

+

+


429

susceptible to proteolytic degradation; however, both a- and P-subunits are eventually degraded (14,44) whereas the y-subunit is more resistant. Proteolytic activation eliminates binding by extrinsic calmodulin (37, 141) but increases sensitivity to Ca2+ activation via the intrinsic &subunit (106).From crosslinking studies, it can be shown to induce the same conformational rearrangements in the P-subunits, as observed also with activation either by phosphorylation or high pH (83). That a-subunit proteolysis results in P-subunit conformational shifts, enhanced activity (presumably of the y-subunit), and modified binding of Ca2+ by the 6-subunits, shows that there must be a high degree of interaction between all of the subunits of this complex molecule. Most likely, proteolytic activation is of little or no physiological significance because of its irreversibility. However, it does pose a major problem to the maintenance of native enzyme in crude extracts or during purification. Even minor proteolytic "nicking' gives rise to significant erroneous results as emphasized in particular by the evidence that proteolyzed phosphorylase kinase requires lower Ca2+ concentrations for activation (54,106).As noted, however, despite proteolytic nicking, the basic structure and mass of the holoenzyme can remain intact (3%). Quite possibly, the results of several of the past studies of phosphorylase kinase may have been compromised because the enzyme used was modified during purification. In general this has been monitored by an assessment of the activity ratio at pH 6.8 to that at pH 8.2, but even this may not be a sufficient criterion (106).It appears important that in preparing phosphorylase kinase one should attempt to achieve a stoichiometry of a : P subunits; in most preparations, actual unity of these subunits is rarely reported.

VII. Covalent Regulation: In Vitro Studies A. INTRINSICPHOSPHATE CONTENT Nonactivated phosphorylase kinase contains intrinsic alkali-labile phosphate, although the actual amount present is in some dispute. Three reports provide 7.18 k 0.95 mol(142), and 7.84 k 1.32 mol(143) per values of 8.5 mol (141), ( a P ~ 6 )but ~ ;in an additional study from one of these same laboratories (27) the amount was reported as 19.4 k 0.7 mol per mol, with the a-subunit containing 2.7 ? 0.4 mol/mol subunit, the P-subunit 1.9 k 0.5 mol/mol subunit and the yand &subunits less than 0.3 mol/mol subunit. From the latter study, it appears that skeletal muscle phosphorylase kinase can be isolated at one of two levels of phosphorylation; the difference between the two forms or the reason for differences obtained using apparently the same purification procedure are not known. The function of this intrinsic phosphate is also not known. It is not present in

430

CHERYL A. PICKE'lT-GIES AND DONAL WALSH

the sites phosphorylated by the CAMP-dependent protein kinase (142), cannot be removed by alkaline phosphatase (143) or protein phosphatase 1 or 2 (143, 144), does not appear to contribute to the activity of the enzyme, and if it does turn over in the cell must do so with a half-life of at least many hours since prolonged incubation (6 h) of intact isolated muscle with 32P-inorganicphosphate does not lead to such levels of intrinsic labeling (145).

B . PHOSPHORYLATION: GENERALCONSIDERATIONS Phosphorylation of phosphorylase kinase with accompanying changes in its activity can be catalyzed in vitro by a number of different enzymes. These include the CAMP-dependentprotein kinase (13, 1 4 , phosphorylase kinase itself [i.e., Ca2 -dependent autophosphorylation (831, the cGMP-dependent protein kinase (146, 147), a Ca2+-calmodulin-dependent protein kinase (148), and a Ca2 - and CAMP-independent protein kinase (glycogen synthase kinase- 1) (149, 1 4 9 ~ )Initial . reports that it was also phosphorylated by a Ca2+-phospholipid-dependent protein kinase (or Ca2+-dependent protease-activated kinase) (150, 151) are probably incorrect (152). The phosphorylations catalyzed by the CAMP-dependentprotein kinase and by autophosphorylation have been the most thoroughly studied and are given most attention here. Both p- and a-or a'-subunits of phosphorylase kinase are phosphorylated by these enzymes. The two reactions can be differentiated and thus each can be studied fairly exclusively due to several differences in their catalytic properties. First, the autophosphorylation process requires Ca2 where the CAMP-dependent protein kinase operates independently of this ion (55, 87). Second, autocatalysis, unlike the protein kinase-catalyzed reaction, occurs more rapidly at pH 8.2 than at pH 6.8 (7). One further means of differentiating the two processes is in the use of the inhibitor protein for the CAMP-dependent protein kinase, which, though a potent inhibitor of this enzyme, has no effect on phosphorylase kinase autophosphorylation (87).For some time it was thought that the MgATP2- requirement for autophosphorylation was much greater than that for the protein kinase, so that this was an additional means of separating the two reactions (87). Subsequent studies (discussed in Section VII,D), however, indicate that the two reactions have a similar K,,,for the nucleotide (18, 88). Mg2+ does stimulate autophosphorylation (67) and this may explain observations that led to the earlier conclusion (87). However, since Mg2+ can also affect the kinetic parameters of the protein kinase-catalyzed reaction (153), and because it appears to have an effect on phosphorylase kinase as a substrate ( 1 5 4 , it is doubtful whether this ion can be used as a tool in differentiating the two reactions. Equally troublesome, different effects of Mg2 on the CAMP-dependent protein kinase-catalyzed phosphorylation have been reported from different laboratories (152, 154), and divalent cation effects on phosphorylase kinase phos+

+

+

+


43 1

phorylation appear to be different for the different isozyme types (21, 142, 144, 154). C. CYCLICAMP-DEPENDENT PROTEIN REACTIONS KINASE-CATALYZED The in v i m phosphorylation of skeletal-muscle phosphorylase kinase by the CAMP-dependentprotein kinase has been studied by several groups of investigators (13, 14, 142, 154). Phosphorylation of both P- and a-subunits has been consistently observed, as has the temporal pattern in which P-subunit phosphorylation is initiated prior to that of the a-subunit and proceeds at a much faster rate. We have presented an extensive characterization of the protein kinase and phosphorylase kinase concentration dependencies of the subunit phosphorylation (142). Part of these data are depicted in Fig. 4. These data are compatible with and extend previously published results (13, 14, 21, 154), and, taken as a whole, permit the following conclusions: 1. The rates of phosphorylation of either subunit show the expected variation with protein kinase concentration (i.e., they increase in an essentially linear manner with increasing protein kinase concentration (for example, examine the panels vertically in Fig. 4). 2. Increasing the concentration of phosphorylase kinase at a fixed concentration of protein kinase (examine the panels horizontally in Fig. 4) produces the expected increase in the amount of time required to achieve a given stoichiometry of phosphorylation, but a simple Michaelis-Menten relationship is not obeyed. 3. The phosphorylation of the P-subunit always appears to be initiated immediately, and even when rates are low there is no indication of an initial lag. 4. The phosphorylation of the P-subunit always precedes that of the a-subunit and a lag in a-subunit phosphorylation is often apparent. It appears as though asubunit phosphorylation does not occur until the P-subunit has been phosphorylated to a level of at least 1 mol/mol (apys),. 5 . The phosphorylation of the a-subunit is catalyzed directly by the CAMPdependent protein kinase. An alternative possibility that could be envisioned from inspection of the progress curves would be that (a) the CAMP-dependent protein kinase catalyzed the phosphorylation of the P-subunit, .(b)this leads to protein kinase catalyzed phosphorylation of the (3-subunit, and (c) enhanced phosphorylase kinase activity results in an increase in a-subunit phosphorylation via autophosphorylation (i.e., a-subunit phosphorylation catalyzed by phosphorylase kinase). This alternative has been eliminated by studies showing that addition of the inhibitor protein of the CAMP-dependent protein kinase, after Psubunit phosphorylation is maximal, blocks further a-subunit phosphorylation (142).

0

Subunit Phosphorylation (mol phosphate incorporated/mol of enzyme) N W P n 0 0 0 0 0 0 0 0

-

0

"

g

o

a Subunit

{ 0,9 Subunit N

0

0

0

W

0 1

P

0

-t

0

-

-

0

0

8

8

-

0

8

-

>

-

w o

N

0 0 0 0 0 Phosphorylase Kinase Activation (increase in units of octivity/mg of phosphorylase kinase) Subunit Phosphorylotion (mol phosphate incorporated /mot of enzyme 1 0 -

N

g

go

6

g

g

\\

0

-

0

Phosphorylase Kinase Activation (increase in units of activity /mg of phosphorylase kinase 1

I Ir

0

g

433


6. The maximal rate of @-subunitphosphorylation is 5- to 10-fold greater than the maximum rate of a-subunit phosphorylation (142, 155). It appears from the various data that the initial phosphorylation of at least one @-subunitis required to produce a conformational change that then permits asubunit phosphorylation. A speculative possibility from these in v i m data is that the nature of the conformational change is similar to that seen upon preincubation with Mg2+ and Ca2+ (see Section V,B), since both apparently require the same period of time to be manifested. One reason to make this proposal is that there does not appear to be a clear relationship between the time courses of @-subunit and a-subunit phosphorylation. Thus, it might appear that the sequence of events is as shown in Scheme LI. P-Subunit phosphorylation

~

conformational shift

~

a-subunit phosphorylation

SCHEME I1

where the conformational shift is rate limiting. Observations on the stoichiometry of subunit phosphorylation by the CAMPdependent protein kinase have been somewhat discrepant. In early studies with the skeletal muscle enzyme, Hayakawa et al. (13) observed a maximum incorporation of 1.8 mol of phosphate/mol p4 and in excess of 4 mol/mol a 4 , while Cohen (14)reported that phosphorylation plateaued after one phosphate had been incorporated into each a- and @-subunit. These latter values are those most apparently accepted, although there remains some uncertainty since the reports specifically documenting a stoichiometry of 4 mol of phosphate into the @have been few. In a subsequent study of ours (142), the subunit/mol of maximum level of @-subunitphosphorylation ranged from 2.5 to 3.2 mol/mol of (a@-ys), with an average of 3.03 t 0.27 mol/mol of (aP$), (n = 25). In this latter study (142) a number of parameters that might have resulted in an aberrantly low value were specifically excluded. The same extent of phosphorylation was obtained whether assayed by 32Piincorporation or colorimetric measurement of total phosphate. An impairment of the conditions of phosphorylation occurring during the reaction was ruled out since addition of more substrate (phosphorylase kinase) resulted in its ready phosphorylation. Since nonactivated phosphorylase kinase contains intrinsic phosphate (see Section VII,A) a less than stoichiometric phosphorylation would be obtained if some of the specific sites in FIG. 4. Time courses of subunit phosphorylation and activation of phosphorylase kinase with varying concentrations of CAMP-dependent protein kinase and phosphorylase kinase. Phosphorylase kinase was phosphorylated under the conditions given in Ref. (142) with 3.9 [panels (a), (b), (c)], 39 [panels (d), (e), (01,or 390 [panels (g), (h), (i)] punits/ml of protein kinase, at phosphorylase kinase concentrations of 0.12 [panels (a), (d), (g)], 0.72 [panels (b), (e), (h)] or 4.32 [panels (c), (0, (i)] mglml. From Pickett-Gies and Walsh (142) with permission.

434


the nonactivated enzyme already contained phosphate. This possibility was eliminated by pretreatment of the enzyme with the appropriate protein phosphatase to dephosphorylate such sites, which did not change the observed degree of phosphorylation. Accurate determinations of stoichiometry are, however, difficult because the calculated value relies on the accurate measurement of several parameters. Nevertheless, from the available data there is clearly some doubt as to whether fully maximal P-subunit phosphorylation (i.e., 4 mol/(aPyS),) of the skeletal-muscle a isozyme readily occurs. In this regard, the data with cardiac phosphorylase kinase are of particular interest. The phosphorylation of bovine cardiac phosphorylase kinase (a'isozyme) has been studied under a variety of conditions (22, 144, 156). The basic pattern of phosphorylation is consistent with that of the rabbit skeletal-muscle a isozyme; namely, P-subunit phosphorylation precedes that of the &'-subunit and occurs at a faster rate. However, the maximal level of phosphate incorporation into the P-subunit is only 1 mol/mol p4. Appropriate controls of the type described above for the skeletal-muscle enzyme have shown that this measured level is not aberrant. With the cardiac-muscle enzyme, the magnitude of difference between the observed level of phosphorylation [ l mol/mol of ((.~f3yS)~] and the phosphorylation of all four P-subunits is such that it can definitely be concluded that the latter does not readily occur. Thus within the cardiac enzyme, interactions must exist of a type akin to negative allosterism whereby, once one of the four (hubunits is phosphorylated, a conformational change must occur that blocks the phosphorylation of the same peptide site in the other P-subunits. Given this conclusion for the cardiac-muscle isozyme, the argument becomes much more tenable that in skeletal-muscle phosphorylase kinase, fully maximal 6-subunit phosphorylation [i.e., to 4 mol/mol (cxP~S),] is significantly inhibited once three of the four sites have been phosphorylated. Of note, the a' isozyme from bovine skeletal muscle gave different results from those obtained for the cardiac enzyme with the f3-subunit being phosphorylated by the CAMP-dependent protein kinase to a level of 3.2 mol/mol (aPy6),(21). Possibly this difference is seen because, although both enzymes are designated as a' isozymes, they are not identical; differences between a isozymes and between a' isozymes have been noted (64).One reason why phosphorylase kinase may not be stoichiometrically phosphorylated could be that the four P-subunits in a molecule of enzyme may not in fact, be identical. If the differences were minor, like only a few amino acid substitutions, the techniques used would not distinguish differences. From what is presented here, an additional speculation arises. It appears that the phosphorylation of the first (of the four) P-subunits in phosphorylase kinase markedly affects the enzyme's conformation. With both the cardiac (a')and skeletal muscle (a)isozymes, the first mole of phosphate incorporated into the P subunits appears essential in order for a subunit phosphorylation to occur. In the

435


cardiac enzyme this also blocks the phosphorylation of the other three @ subunits. It is tempting to speculate that the same conformational change directs the subsequent site of phosphorylation; namely, to depress P-subunit phosphorylation but enhance that of the a subunit. Most generally, CAMP-dependent protein kinase-catalyzed a-subunit phosphorylation achieves a stoichiometry of 4 mol/mol of (aP$),, or somewhat higher. This is observed with either the a isozyme from rabbit skeletal muscle (13, 14, 142, 154) or the a’ isozyme from either bovine heart (144, 156) or bovine red-skeletal muscle (21). Singh and Wang (154) studied the effects of Mg2 on the extent of a subunit phosphorylation and showed that by increasing the Mg2 from 1 mM to 10 mM there was an increase in the extent of a-subunit phosphorylation from 4 to 20 mol/mol of (aPy8),. Thus Mg2+ in this concentration range caused more a sites to become available for phosphorylation, presumably by interacting with Mg2 -specific sites on phosphorylase kinase (see Section V,B). Under one restricted set of conditions we have found evidence that, as previously described for the P-subunit, interactions between a-subunits can limit the degree of their phosphorylation (142). If the free Mg2 concentration is reduced to 20 pM, the phosphorylation of the P-subunit is normal but that of the a-subunit is only to the level of 2 mol/mol of (a@y8),.As with P-subunit phosphorylation, this reduced level of a-subunit phosphorylation is not the result of conditions of phosphorylation becoming impaired. It appears again to be indicative of a negative allosteric-type effect whereby the initial phosphorylation of two sites leads to a conformational change in the protein that blocks subsequent phosphorylation of identical sites on other a-subunits. Presumably, Mg2 binding interferes with this conformational change. It is apparent that the phosphorylation of phosphorylase kinase produces a sequence of incremental changes in the conformational organization of the molecule. To what extent these properties of the isolated enzyme pertain to the enzyme in the physiological milieu remains an important question to be addressed. There are also distinctions between the isozyme types; the principal differences documented are in the level of P-subunit phosphorylation and in the effects of Mg2 . Mg2 increases the level of a-subunit phosphorylation of the a isozyme. In contrast, Mg2+ inhibits a’subunit phosphorylation of either the cardiac enzyme (156) or the a’ isozyme from red-skeletal muscle (21). Whether these various dissimilarities are indicative of actual regulatory differences in the isozymes remains to be evaluated. Factors that might modify the CAMP-dependent protein kinase-catalyzed phosphorylation of phosphorylase kinase remain for the large part unexplored. It appears particularly important to examine those factors with which phosphorylase kinase is known to interact within the cell, notably, phosphorylase, calmodulin, glycogen, and the other constituents of the glycogen particle. Cox and Edstrom (157) reported that extrinsic calmodulin inhibits CAMP-dependent protein kinase-catalyzed phosphorylation of the @-subunitto the extent that, in +

+

+

+

+

+

+

436

CHERYL A. PICKETT-GIES A N D DONAL W A L S H

the presence of calmodulin, the rates of a-and P-subunit phosphorylation were very similar. In the isolated glycogen particle, the CAMP-dependent phosphorylation of phosphorylase kinase can be readily demonstrated (158). Both aand P-subunits are phosphorylated, reaching stoichiometries similar to those observed with the isolated enzyme, and showing the same temporal pattern, namely, P-subunit phosphorylation is faster and thus appears to precede that of the a-subunit . These subunit phosphorylations are EGTA-insensitive and blocked by addition of the inhibitor protein of the CAMP-dependent protein kinase. They are thus not a consequence of Ca2 -dependent activation as occurs in the flash activation of phosphorylase in these particles (97) (However, autophosphorylation can also be demonstrated). Other parameters of these reactions in the glycogen particle, particularly effects of phosphorylase and glycogen, remain to be explored. Singh and Huang (158) have reported that CAMP-dependent phosphorylation of both the a and P subunits was selectively activated by spermine (10 pA4) or spermidine (150 I-LM), although other substrates for the CAMP-dependent enzyme were not affected. The sequences of the specific phosphorylation sites of rabbit skeletal-muscle phosphorylase kinase have been identified by Cohen's laboratory (155, 159) as shown below: +

P-Subunit: Val

Ala-Arg-Thr-Lys-Arg-Ser-Gly-Ser(P)-Ile-Tyr-Glu-Pro-Leu-Lys-Ile u-Subunit:

Phe-Arg-Arg-Leu-Ser( P)-Ile-Ser-Thr-Glu-Ser-Gly SCHEME I11

The Val or Ile of the P-subunit sequence represent two alleles. The sequences for both the a-and P-subunits contain the pair of basic amino acids N-terminal to the target serine that are considered one of the major recognition sequences dictating substrate specificity of the CAMP-dependent protein kinase. It is of interest that for the P-subunit one of the basic amino acids is lysine rather than arginine, and there are two rather than one intervening amino acids. With artificial peptides both of these changes diminish the affinity for the substrate; however, in the inhibitor protein of the CAMP-dependent protein kinase, an arginine located on the N-terminal side of the basic subsite makes a marked difference in inhibitory potency (1594. The sites phosphorylated on the a' and P subunits of the a' isozyme have not yet been characterized. The phosphorylation of phosphorylase kinase results in its activation but understanding of the relationship between the two is incomplete. Maximum phosphorylation leads to up to a 50-fold increase in activity at pH 6.8 with either the skeletal muscle a isozyme (7) or a' isozyme (21), but only a 2- to 3-fold increase with the cardiac a' isozyme (144). These increases are due primarily to an

437


increase in affinity for phosphorylase, as is also seen in activation by proteolysis or allosterically by Ca2 . As previously described, there are also some changes +

in the requirements for Ca2 and Mg2 . Phosphorylation by the CAMP-dependent protein kinase results in a conformational change that is similar to that which occurs with activation by proteolysis, increased pH, or Ca2 plus Mg2 ; this is evidenced by cross-linking studies that showed a greater proportion of P-P dimers formed with enzyme activated by any of these ways (38). There has been a lack of concurrence about the molecular events that result in activation. A thesis presented by Cohen (14), which was based on activation and inactivation profiles (under one set of reaction conditions), stated that activation is the result of P-subunit phosphorylation alone. Later, additional support for this was provided by a comparison of activation by the CAMP-dependent and cGMPdependent protein kinases (147). Each phosphorylated both subunits, but with different ratios of activities towards them. Equal increments of activation, however, were correlated with P-subunit phosphorylation but not a-subunit phosphorylation. Concomitant with the above proposal presented by Cohen ( I # ) , Hayakawa et af. (13) reported, in contrast, that a correlation was not observed between phosphorylation of either subunit and enzyme activation; rather it was concluded that the phosphorylation of both contributed to the increase in activity. Most subsequent data supports this latter position. Singh and Wang (154), for example, described a biphasic pattern of phosphorylation and activation at high Mg2+ concentrations. Under these conditions an initial rapid rise in activity coincided with either P- or both a-and P-subunit phosphorylation, followed by a slow increase in activity that corresponded to further a-subunit phosphorylation. A similar biphasic pattern was also observed with the a’isozyme from either bovine cardiac (144) or bovine red-skeletal muscle (21) (independent of the Mg2+ concentration). With either, the first phase of activation appears to be best correlated with P-subunit phosphorylation, but a second slower phase then occurs after P-subunit phosphorylation is maximal, and this is clearly dependent upon CAMP-dependent a-subunit phosphorylation. Especially for the cardiac enzyme, where P-subunit phosphorylation occurs only to the level of 1 mol/mol of (ap-ys),,the second a-subunit phosphorylation-dependentphase of activation is marked. Also with the cardiac enzyme, inactivation has also been shown to be correlated with a-subunit dephosphorylation using an enzyme preparation where the P-subunit phosphate was stabilized as the thiophosphate (160). a-Subunitdependent activation of the skeletal muscle a isozyme is well-exemplified by the studies of Singh et af. (149, 149a). As discussed in more detail later, phosphorylation was examined using two enzymes, the CAMP-dependent protein kinase, and a Ca2 and CAMP-independent enzyme which phosphorylated the P-subunit in the same site as the CAMP-dependent protein kinase, but did not phosphorylate the a-subunit. Phosphorylation with this latter enzyme activated phosphorylase kinase, but not maximally. The activation was clearly due to P+

+

+

+

+

438


subunit phosphorylation. Following this treatment, addition of the CAMP-dependent protein kinase resulted in no further phosphorylation of the P-subunit, but the a-subunit was phosphorylated and the enzyme further activated. Thus, this latter activation is clearly dependent on a-subunit phosphorylation. Data obtained with the isolated glycogen particle also show that both a- and P-subunit phosphorylation contribute to activation ( 7 5 ~ ) . Expanding upon this data, we have further examined the relationship between subunit phosphorylation and activation, over a range of protein kinase and phosphorylase kinase concentrations, previously commented on (Fig. 4) in reference to concentration dependency of subunit phosphorylation. The general trends for activity changes are as follows: 1. The rates of activation show the expected variation with protein concentration and at each fixed concentration of phosphorylase kinase increase approximately linearly (examine panels vertically, Fig. 4). 2. The rates of activation with varying substrate concentration do not obey a simple Michaelis-Menten relationship but at the higher phosphorylase kinase concentration increase markedly (examine panels horizontally, Fig. 4).

3. No simple correlation is apparent between enzyme activation and subunit phosphorylation. Some conditions show data similar to that first presented by Cohen (14), namely an apparent correlation with @-subunit phosphorylation alone, whereas others more closely mirror the results of Hayakawa et al. (13) with the phosphorylation of both subunits apparently contributing to the increase in activity. The data available indicate that the relationship between subunit phosphorylation and enzyme activation is clearly complex. The potential pathways for conversion of a4@,y48, to (a-P),(P-P),y,G, are illustrated by Fig. 5 . Minimally, in the conversion of a4P4to (a-P),(@-P),, seven partially phosphorylated intermediates are generated. Maximally, there is a possibility of 23 intermediates, 40 possible reactions involving unique substrate species, and at least 70 possible routes to go from a4P4to (a-P),(P-P),. Although it is unlikely that all of these are actually utilized, this degree of possible complexity might explain why no simple correlation between subunit phosphorylation and activation is apparent from activation studies. It might be anticipated that the various phospho intermediates will not behave identically as substrates, and this clearly appears to be the case. Thus, as an example, with the a isozyme, phosphorylation of the asubunits appears to require the prior phosphorylation of at least one P-subunit, but whether subsequent P-subunit phosphorylation further affects the reaction rate of a-subunit phosphorylation is not known. A second question to be considered is whether the incremental phosphorylation of a single subunit produces equal increments in activity. As the reaction from a4P4to (a-P),(p-P), proceeds


- - I I 1- 1- I

a4

a4

a4

P4

P3 P- p

P2

a3

a-P

P4

~

(P-PI2

a4

04

-(P-P).q

(P-PI,

a3

a3

a3

a3

a-P

a-P

a-P

P2

P

a-P (P-PI4

P3

439

(P-P),

I

I

(a-P),

(U-P),

(a-P),

P4

P3 P- P

(P-PI,

(a-PI4

(U-P),

P2

I -I -I -I- I

(a-P),

P4

P3

P2

P-p

(P-P),

(a-P)4

P

(Q-p)4

-(P-P)4

(P-P),

FIG. 5. Phospho intermediates in the activation pathway of phosphorylase kinase (y-and 6subunits are not indicated). From Pickett-Gies and Walsh (142) with permission.

different intermediates will be generated, each of which may next be phosphorylated in either the a-or @-subunit.Unless phosphorylation of one or another were to be exclusively favored, which does not appear likely from available data, the subsequent phosphorylation event would lead to multiple products. This would lead, as the reaction progressed beyond the phosphorylation of the first P-subunit, to the eventual generation of many of the species depicted in Fig. 5 . Quite possibly, which product is formed may depend upon the concentrations of both the enzyme and the substrate. If incremental phosphorylation of a site were not to lead to incremental increases in activity, this could well explain the array of apparent correlations depicted by the type of data presented in Fig. 4. As previously discussed, there are several lines of evidence that have implicated the phosphorylation of both the a- and P-subunits in the regulation of activity. In what appeared to be in contrast to this, Ganapathi and Lee (161) showed with dephosphorylation of skeletal-muscle phosphorylase kinase (phosphorylated in both a-and @-subunits)that a correlation is observed between p-

440

CHERYL A. PICKETT-GES AND DONAL WALSH

subunit dephosphorylation and enzyme inactivation, and that the enzyme can be fully inactivated while still containing phosphate in the a-subunit. One possibility from this data, however, was that activation required p subunit phosphorylation, but could be further enhanced by phosphorylation of the 01 subunit, even though a subunit phosphorylation alone does not cause activation. This is what we have now shown to be the case (161a, 161b). The dephosphorylation of phosphorylase kinase has been studied using two highly specific protein phosphatase preparations: one (the catalytic subunit of the ATP Mg-dependent protein phosphatase) selective for the p subunit, the other (the catalytic subunit of the polycation dependent protein phosphatase) specific for the site(s) on the a subunit phosphorylated by the CAMP-dependent protein kinase. Using this approach it is seen (161a, 161b) that as Ganapathi and Lee (161) reported there is a close linear correlation between P subunit dephosphorylationand enzyme inactivation; further, enzyme fully dephosphorylated in the p subunit is fully inactivated. 01Subunit phosphorylation, however, also regulates activity and does so by linearly amplifying the effect of p subunit phosphorylation. Thus, enzyme phosphorylated in both the a and p subunits is inactivated as a consequence of selective 01 subunit dephosphorylation but the extent of inactivation is dependent on the level of p subunit phosphorylation. Thus, complete a subunit dephosphorylation does not fully inactivate the enzyme. As discussed earlier, with phosphorylation the incorporation of the first mole of phosphate into the p subunit appears to result in a conformational change that permits a subunit phosphorylation. Possibly, it is this same conformational change that permits subsequent 01 subunit phosphorylation to modify activity, and without this phospho-P-subunit induced conformational change, a subunit phosphorylation is without effect. The reason for the regulation of phosphorylase kinase by phosphorylation in different sites but by the same enzyme, and presumably in response to the same stimuli, is obscure. One possibility, suggested by several investigators, is that different characteristics or functions might be regulated by phosphorylation of different sites. Phosphorylation of phosphorylase kinase is known to enhance activity toward glycogen synthase (88, 89) and phosphorylase kinase itself (87); however, the relationship between specific subunit phosphorylation and activity toward these substrates has not been explored. One early suggestion by Cohen and Antoniw (162) was that a-subunit phosphorylation regulated the dephosphorylation of the P-subunit. This proposal, termed by the authors “second site” regulation, was based upon observations of dephosphorylation catalyzed by a contaminating phosphatase. Later experiments, however, failed to reproduce the initial findings (163), and both Ganapathi et al. (164) and ourselves (161b), specifically exploring this possibility with purified enzymes, have found no evidence that the phosphorylation of one subunit modifies the rate of dephosphorylation of the others. The CAMP-dependent phosphorylation and concomitant activation of phos-

44 1


phorylase kinase has also been explored in the isolated glycogen particle (7%). The glycogen particle-associated enzyme can be phosphorylated either by endogenous CAMP-dependent protein kinase (as a consequence of exogenously added or endogenously produced CAMP) or by exogenous addition of protein kinase catalytic subunit. The characteristics of all three are similar with the phosphorylation of p subunit slightly proceeding that of the a subunit and with associated enzyme activation. Examining correlations between subunit phosphorylation and enzyme activation was not practical because of concomitant Ca2 -Mg2 -dependent synergistic activation that occurred as the ATP was consumed. Clearly from these studies, however, phosphorylase kinase in the glycogen particle is equally accessible to intrinsic or extrinsic CAMP-dependent protein kinase. Also evident from these studies were the presence in the glycogen particle of both p subunit and a subunit phosphatases, with the latter being Ca2 -dependent. +

+

+

D. AUTOPHOSPHORYLATION Phosphorylase kinase can catalyze its own phosphorylation in the presence of MgATP2- and Ca2+ (8, 61, 62, 87). This is not a novel property of phosphorylase kinase as many protein kinases appear to be capable of autophosphorylation although the physiological significance of this is not known. Autocatalysis results in the incorporation of phosphate into both a- and psubunits with an accompanying increase in enzyme activity toward phosphorylase b. Activation can be as much as 80-to 100-fold (depending on conditions of the reaction) and thus of greater magnitude than that generally observed upon phosphorylation by the CAMP-dependent protein kinase (18). Using the partially dissociated complexes, Chan and Graves (45) observed that the y8 complex did not autophosphorylate, but did catalyze EGTA-insensitive phosphorylation and activation of the holoenzyme. The ay8 complex did autophosphorylate, but its activity was unaffected. The mechanism of autophosphorylation is not yet clearly understood. Studies by King et al. (88)and Hallenbeck and Walsh (18) have reported that the K,,, for MgATP2- is in the range of 17-27 pJ4 (a value quite similar to the nucleotide requirement of the CAMP-dependent protein kinase). This MgATP2- requirement is somewhat lower than the reported K,,, range for the phosphorylase b to a reaction (i.e., K,,, = 70-240 pM). The K , for Mg2+ in the autocatalytic reaction also differs from that in the phosphorylase b to a reaction, being approximately 6-fold higher (18). Mg2+, in this case, may be affecting not only the catalytic activity of phosphorylase kinase but its ability to serve as a substrate as well. Support for this suggestion can be found in the observation that Mg2 increases the extent of phosphorylation by the CAMP-dependent protein kinase (154). It has been suggested by Carlson and Graves (51), that autophosphorylation and the +

442


phosphorylase b to a reaction may be catalyzed by different catalytic sites on the enzyme. These investigators observed that phosphorylase b and peptide analogs stimulated, rather than inhibited, autophosphorylation. Whether or not the differences in kinetic parameters indeed are a further reflection of different catalytic sites is unclear. Autophosphorylation could potentially be either an intra- or intermolecular process. Two groups of investigators have reported that the initial velocity of autophosphorylation is linear (i.e., the reaction is first-order) with respect to enzyme concentration (18, 88). This observation is consistent with an intramolecular, but not with an intermolecular, process. Interestingly, in earlier studies, DeLange et al. (8) reported that the activation of nonactivated kinase was stimulated by the addition of activated kinase, and Chan and Graves (45) have shown that the y8 partial complex can phosphorylate the holoenzyme. These latter observations appear to be more consistent with an intermolecular mechanism of autophosphorylation. Whether or not both mechanisms can occur is not clear. The pattern and extent of subunit phosphorylation and activation observed during the autocatalytic reaction have, like those of the protein kinase reaction, varied considerably. This appears to be due at least in part to the dependence of this process upon buffers, pH, and divalent cations. At a pH near neutrality there appears to be a lag in both a- and p-subunit phosphorylation as well as in activation (51, 61). This lag is particularly pronounced in glycerophosphate buffer, which appears to inhibit autophosphorylation (as do phosphate and several phosphate-containing compounds) (61). At a pH of 8.2, or at pH 6.8 following preincubation with Mg2+ plus Ca2+ (88),this lag is not observed. Regardless of pH, the phosphorylation of the a- and P-subunits appears to commence simultaneously (i.e., a-subunit phosphorylation does not lag behind that of the psubunit as it does in the protein kinase-catalyzed reaction) (18, 61, 88). The maximal levels of autophosphorylation reported have varied from 1 to 4 mol phosphatehol p and from 3 to greater than 5 mol phosphatehol a. While this may suggest that the potential number of sites that can be autophosphorylated is much greater than that which can be phosphorylated by the CAMPdependent protein kinase, in general autophosphorylationis carried out at higher Mg2+ concentrations than the latter reaction. Singh and Wang (154) have reported that the autocatalytic process and the protein kinase-catalyzed reaction at high Mg2+ are similar with respect to both the final level of phosphorylation and the activation attained. Discrepancies also exist as to the correlation between subunit phosphorylation and activation. Wang et af. (61) and Hallenbeck and Walsh (28),who carried out autophosphorylation in buffers near neutrality, observed coincident phosphorylation of both a- and p-subunits. Activation that occurred comcomitantly with phosphorylation could not be correlated with the phosphorylation of a specific


443

subunit. In contrast, King et al. (88),performing experiments at pH 8.0 reported that activation correlated well with P-subunit phosphorylation. In these experiments, P-subunit phosphorylation clearly reached a maximum prior to the complete phosphorylation of the a-subunits. Activation also plateaued before asubunit phosphorylation did. Chan and Graves (45) have shown that autophosphorylation of the partial complex (ayS) results in the incorporation of up to 4 mol phosphate per mol without any effect on catalytic activity. This might suggest that autophosphorylationof the a-subunit does not affect activity; however, the ayS complex has a higher specific activity (at pH 6.8) than the holoenzyme, so possibly it is already in a conformational state equivalent to that promoted by autophosphorylationin the holoenzyme and thus phosphorylation is without further effect. Indeed, a role for a-subunit phosphorylation in autoactivation has been implicated in other experiments. Hallenbeck and Walsh (18) found that when MnATP2-, rather than MgATP2-, was used in autocatalytic reactions, a- and P-subunit phosphorylation were no longer coincident, and p phosphorylation plateaued at 1 mol of phosphate/mol P. Under these conditions, both activation and a-subunit phosphorylation continued in the absence of additional P-subunit modification. In another study, Sul et al. (156) observed an increase in CAMP-independent a' subunit phosphorylation (most likely autophosphorylation)subsequent to phosphorylation by the CAMP-dependent protein kinase. Concomitant with this increase in a' phosphorylation was an increase in enzyme activity. Taken together the existing data suggest that autophosphorylation of both a-(or a')and P-subunits can alter enzyme activity. Clearly, multiple sites can be phosphorylated on both subunits by this process; it is not clear whether all or only some of these sites control activity. As autophosphorylation and protein kinase-catalyzed phosphorylation have many characteristics in common, the question of whether common sites are phosphorylated by the two enzymes is of interest. This question was approached indirectly in the studies of Wang et al. (61) who observed that when phosphorylase kinase was phosphorylated to a maximal level by either the CAMPdependent protein kinase or by autophosphorylation, and subsequently subjected to the alternate process of phosphorylation, additional phosphate was incorporated. In addition, autophosphorylation of enzyme previously phosphorylated by the protein kinase was accompanied by a further increase in enzyme activity. This suggests that the two mechanisms of phosphorylation do not involve common phosphorylation sites. However, in reactions catalyzed by CAMP-dependent protein kinase at high Mg2+ concentration, Singh and Wang (154) observed a maximal level of phosphate incorporation quite similar to that attained by the autocatalytic process. When both reactions were allowed to proceed simultaneously the same maximal level of incorporation was attained as with either reaction alone. Clearly, conclusive evidence as to the identity of sites phosphorylated by these two mechanisms awaits characterization of all of the phosphory-

444

CHERYL A. PICKE’lT-GIES AND DONAL WALSH

lated sequences. King ef al. (88) have mapped tryptic phosphopeptides derived from phosphorylase kinase that had been autophosphorylatedto slightly less than 1 mollmol P. Two major phosphopeptides were obtained which appeared to be identical to the two allelic phosphopeptides generated from the protein kinasephosphorylated P-subunit. Therefore, it appears that the P-subunit site phosphorylated by the CAMP-dependent protein kinase can also be phosphorylated by autocatalysis. Information on the other sites phosphorylated by these two processes is still needed. The physiological significanceof phosphorylase kinase autophosphorylationis not known. As discussed in Section VIII,B, the bulk of in vivo data suggest that activation by neural or electrically stimulated Ca2+ release is not due to covalent modification of the enzyme. The possibility that autophosphorylation is enhanced subsequent to phosphorylation by the CAMP-dependent protein kinase, however, cannot be ruled out, particularly since this appears to occur in vifro. The observation that glycogen and phosphorylase appear to stimulate autophosphorylation, coupled with their apparent close association with phosphorylase kinase in vivo, suggests that intracellular conditions may permit this type of activation.

E. PHOSPHORYLATION AND ACTIVATION BY OTHERPROTEIN KINASES Many features of the phosphorylation of phosphorylase kinase by other protein kinases have been presented in previous sections of this chapter. The cGMPdependent protein kinase catalyzes the phosphorylation of both the a- and psubunits but in contrast to the CAMP-dependent enzyme, a-subunit phosphorylation is faster than that of the P-subunit and occurs without an initial lag (146, 147). Tryptic phosphopeptides obtained with either protein kinase are identical (in size) but, especially for the a-subunit where the peptide is large (40 amino acids), it is not known if the phosphorylation sites are identical. Cohen (147) has suggested that activation correlates with P-subunit phosphorylation, but the cGMP-dependent protein kinase maximally phosphorylates the p-subunit to a level of only 1 molhol of (aPy8),. The data would fit equally well with the proposal that a-subunit phosphorylation regulates activity only after the P-subunit is phosphorylated. Phosphorylation of phosphorylase kinase by Ca2 and cyclic nucleotide-independent “casein” protein kinase (that is also a glycogen synthase kinase) has been studied by Singh et al, (149, 149a). In the initial report (149) using this enzyme and with a glycerophosphate buffer system the P subunit was observed to be phosphorylated to a level of 1 mol/mol of (aPy8),, and the a subunit only minimally. The P subunit site phosphorylated appears to be identical to that phosphorylated by the CAMP-dependent protein kinase, and likewise causes

-

+

10.

445

PHOSPHORYLASE KINASE

activation. Subsequent phosphorylation of the a subunit by the CAMP-dependent protein kinase, however, resulted in further activation. In a follow-up report using this casein protein kinase (149a) it was observed that in a Tris chloride buffer an incorporation of greater than 2 mol into each of the (Y and p subunits was obtained, but with the prior degree of observed activation. One of the sites on the p subunit that is phosphorylated by the casein kinase is also that phosphorylated by the CAMP-dependent protein kinase and is presumably the site responsible for activation. The sites on the a subunit phosphorylated by the casein kinase and CAMP-dependent protein kinase are distinct, and those phosphorylated by the casein kinase do not appear to affect activity. One of the a sites, however, is only phosphorylated by the casein kinase following prior (Y subunit phosphorylation by the CAMP-dependent enzyme. The role of casein kinase phosphorylation of phosphorylase kinase needs exploration in vivo. Given that phosphorylase kinase as isolated is a phosphoprotein, it is also important to explore whether any of these “endogenous” phosphates are in sites catalyzed by this casein kinase.

F. PHOSPHORYLASE KINASEDEPHOSPHORYLATION A detailed characterization is available of the protein phosphatases capable of catalyzing phosphorylase kinase dephosphorylation. These are only briefly reviewed here, as more detailed descriptions are available elsewhere (165-1 73, 173~2,1736). In general, four protein phosphatases appear to account for most, if not all, of the cellular activity, and are involved in metabolic regulation and phosphorylase kinase dephosphorylation. These have been designated ( 167) Type 1 (ATP Mg-dependent phosphatase) and Types 2A (Polycation-dependent phosphatase), 2B (Calcineurin), and 2C. This designation is based on primary specificity toward either the @-subunit(type 1) or the a-subunit (types 2) of phosphorylase kinase phosphorylated by the cAMP-dependent protein kinase . The type 1 phosphatase is regulated by an inhibitor protein (inhibitor-1), and both it and an intrinsic modulator protein in the phosphatase are regulated by phosphorylation, catalyzed by the CAMP-dependent protein kinase and a Ca2 and cyclic nucleotide-independentprotein kinase, respectively. Type 2B is regulated by Ca2+ and calmodulin. There are no known physiological regulators of types 2A or 2C. Both types 1 and 2A phosphatases have a broad substrate specificity and act upon a range of substrates, whereas 2B and 2C have a much narrower specificity. From studies of their relative cellular amounts, Ingebritsen et al. (173) estimated that the P-subunit is primarily dephosphorylated by the type 1 phosphatase, although in liver type 2A may also contribute to a small degree. Dephosphorylation of the a-subunit appears to depend upon the availability of Ca2+. In the presence of Ca2+, dephosphorylation of the a-subunit would be predominantly catalyzed by type 2B. In the absence of Ca2 , de+

+

446

CHERYL A. PICKEIT-GIES AND DONAL WALSH

phosphorylation, if it occurs, would probably be a consequence of type 2A activity. The effects of dephosphorylationof phosphorylase kinase on its activity and possible other functions are presented in detail in Section VII,C.

G. ADP-RIBOSYLATION Recently (I 73c), the ADP-ribosylation of phosphorylase kinase has been described. Using a hen liver nuclear enzyme, phosphorylase kinase was ribosylated in both the a and p subunits on arginine residues. ADP ribosylation diminished both CAMP-dependent and autocatalytic phosphorylation and, as a result, suppressed phosphorylation-dependent activation. ADP-ribosylation itself did not affect activity.

VIII. Regulation of Phosphorylase Kinase in Intact Cells A. HORMONAL ACTIVATION Since the classical paper by Drummond et af. (IIO),it has been well established that phosphorylase kinase activity in cells is regulated both by Ca2 , allosterically, and in response to CAMP, via CAMP-dependent phosphorylation. The means to distinguish between these two appears straightforward. If changes in the activity state of phosphorylase occur (i.e., the ratio of phosphorylase a to phosphorylase 6 increases) without increases in either the cellular concentration of cAMP or the covalent activation state of phosphorylase kinase, it can, in most circumstances, be reasonably concluded that the mechanism of regulation is via the allosteric, Ca2 -dependent, activation of phosphorylase kinase. In many such situations (as discussed below), further confirmation of the likelihood that Ca2+ is acting as the regulatory agent has come (a) from correlations with what occurs when Ca2+ metabolism is manipulated, such as by omission of external Ca2 or drug-promoted influxes of Ca2 ; (b) from correlations with changes in other Ca2+-mediated processes, such as muscle contraction; and (c) in some situations by direct measurements of internal Ca2 fluxes. Alternatively, regulation of phosphorylase via cAMP is reasonably concluded as the mechanism of control when there are coordinated changes in cAMP levels, CAMP-dependent protein kinase activity ratios, phosphorylase kinase activation state (i.e., covalent phosphorylation state), and phosphorylase a formation. When such data are obtained, the reasonable conclusion is that the covalent activation of phosphorylase kinase is the main contributory cause for phosphorylase activation. A critical question that remains from such initial data, however, is whether or not this is sufficient to explain phosphorylase activation, and whether or not there is +

+

+

+

+

447


an additional need for cellular Ca2 concentrations to be elevated in order for the enhanced activation state of phosphorylase kinase to be expressed. This is discussed in more detail in Section VII1,B. Examination of the regulation of phosphorylase kinase in intact cells has occurred primarily in liver, skeletal muscle, and heart. In all three, both mechanisms of regulation are clearly important. With liver, there is excellent evidence that a-adrenergic agonists, vasopressin, and angiotensin can act specifically via Ca2+. Phosphorylase activation by these three hormones occurs without concomitant changes in cAMP and phosphorylase kinase activation state, whereas it is severely depressed if extracellular Ca2 is omitted. In addition, increasing cytosolic Ca2+ activates phosphorylase in a similar manner. This system, especially with reference to Ca2+, has been extensively characterized. The details are beyond the scope of this review but have been presented elsewhere [see Ref. (174)for extensive references]. In brief, it is apparent that these hormones promote the cellular translocation of Ca2+ leading to an increase in cytosolic Ca2 and, in consequence, phosphorylase activation. The most likely sources of this Ca2+ are partly intracellular (from a specific pool in the endoplasmic reticulum) and partly extracellular (by transport into the cell). At least one of the messengers promoting these changes is inositol triphosphate. The activation of hepatic glycogenolysis by glucagon is clearly distinct from that of the three hormones previously listed. Glucagon-stimulated phosphorylase activation is correlated with increases in cAMP and covalent activation of phosphorylase kinase; thus, glucagon is clearly regulating phosphorylase via activation of the CAMP-dependent cascade. The role of Ca2+ in this cannot be evaluated with certainty because of differences in reported data. Omission of Ca2+ from the medium under conditions that eliminate a-adrenergic phosphorylase activation leads to either no change in glucagon-stimulated activation (175) or a diminished but clearly not abolished effect (113). Most likely the difference in these results is due to differences in the degree to which cellular Ca2+ has been depleted. Under both circumstances it is clear that phosphorylase activation mediated by aadrenergic agonists, vasopressin, or angiotensin (i.e., as catalyzed by nonactivated phosphorylase kinase) has a higher Ca2 concentration requirement than that occurring upon glucagon stimulation (i.e., as catalyzed by covalently activated enzyme). Thus, it appears from the data obtained that covalently activated liver phosphorylase kinase may catalyze a Ca2 -independent activation of phosphorylase, or at least requires a lower concentration of Ca2+ for activity than does the nonactivated form. Most likely, the latter explanation is correct. Intact tissue studies with skeletal muscles show that muscle phosphorylase kinase can likewise be regulated either by Ca2+, allosterically, or by CAMPdependent covalent modification. This was first shown by the classical work of Drummond et al. (110)and has been well borne out subsequently (111, 176, 177). With skeletal muscle, electrical stimulation, either directly or via an in situ +

+

+

+

+

448


nerve, produces rapid formation of phosphorylase a and brisk glycogenolysis, but neither an increase in cAMP nor a covalent activation of phosphorylase kinase. Electrical stimulation also results in contraction; both the contraction and phosphorylase a formation increase with the frequency of stimulation (178, 179), but are diminished by Ca2+ depletion. The stimulus for both is clearly an increase in cytosolic Ca2 , arising primarily by release from the sarcoplasmic reticulum. For muscle contraction, the site of Ca2+ action is troponin C associated with the myofibrils. For the allosteric activation of phosphorylase kinase, three sites are possible, the intrinsic 8-subunit, extrinsic calmodulin, and extrinsic troponin C. Cohen (106)presented cogent arguments in favor of the latter as being the most likely, although this remains to be resolved. The similarity of Ca2 -binding characteristics for the activation of contraction and of glycogenolysis provides for a ready coordination between the two. In contrast to electrical stimulation, P-adrenergic stimulation of skeletal muscle produces a coordinate rise first in CAMP, then in the activation of phosphorylase kinase and, consequentially, in phosphorylase a formation. This clearly implicates activation of the cAMP cascade as the primary route of regulation. However, EGTA addition to the external medium dampens phosphorylase a formation without diminishing catecholamine-stimulated formation of either cAMP or activated phosphorylase kinase (176). Thus, activated phosphorylase kinase still requires Ca2 . Catecholamines, however, do not initiate skeletal muscle contraction; therefore, in contrast to what is observed with electrical stimulation and nonactivated phosphorylase kinase, the concentration of Ca2 required by activated phosphorylase kinase must be lower than the threshold level that would stimulate contraction. This data with intact tissues appears to coincide well with that obtained in v i m with the purified enzyme. As described in Section V,A, the phosphorylation of skeletal-muscle phosphorylase kinase increases its Ca2 affinity; presumably this permits Ca2 to bind at a concentration where Ca2+ is not bound to troponin C in the myofibrils. Two possible situations could be envisioned with the catecholamine stimulation of muscle glycogenolysis; either phosphorylation of phosphorylase kinase lowers its requirements for Ca2+ down to the levels present in ambient muscle, or catecholamines increase cytosolic Ca2 but not to levels sufficient to initiate contraction. Most likely, the latter is correct. Measured levels of ambient Ca2+ appear to be substantially lower than that required for activated phosphorylase kinase, and the manipulations showing that activated phosphorylase kinase in the cell still required Ca2+ (176) would probably not have diminished the ambient free Ca2+ concentration. Further, Stull and Mayer (111) reported that very low concentrations of isoproterenol stimulate phosphorylase a formation without concomitant increases in cAMP or covalent activation of phosphorylase kinase. Most likely, this concentration of isoproterenol acts indirectly by increasing cytosolic Ca2+. Data of Gross and Johnson (176) suggest that isoproterenol +

+

+

+

+

+

+

449


stimulates a Na2 -dependent Ca2 channel across from the transverse tubules, which because of its location may be selectively available to phosphorylase kinase rather than the myofibrils. These types of channels, however, have not been observed by other approaches. In cardiac muscle the regulation of phosphorylase kinase appears similar to that in skeletal muscle, but there are some noted differences. Elevation of external Ca2+ (180, 181), anoxia (182, 183), and a-adrenergic stimulation (73) all cause phosphorylase a formation without changes in either cAMP or phosphorylase kinase activation state; thus for these the likely regulation is via Ca2 stimulation of unactivated phosphorylase kinase. This situation, however, is different from what occurs in skeletal muscle, where Ca2 -stimulated contraction and Ca2+ stimulation of nonactivated phosphorylase kinase (as, for example, with electrical stimulation) appear to be coordinately linked. In contrast, cardiac contraction occurs apparently without concomitant phosphorylase activation and the latter appears to occur either when Ca2 is elevated further or via a different route. This suggests that in cardiac muscle either the two processes are differentially sensitive to Ca2 or are exposed to different pools. An important consideration, however, is not simply whether Ca2+ has been increased to a given level, but how long it has been elevated. There are clearly marked differences in the time constants for activation of contraction, and subsequent relaxation, and for phosphorylation of phosphorylase, and subsequent dephosphorylation. Phosphorylase activity is only meaningfully (i.e., measurably) increased if Ca2 is elevated for a considerably longer time than is necessary to evoke a contractile response. The CAMP-dependent regulation of cardiac phosphorylase kinase has also been well studied. Either P-adrenergic (184-186) or glucagon (187) stimulation with perfused hearts, or P-adrenergic stimulation with either papillary muscle (183)or isolated ventricular strips (188)results in coordinated changes in CAMP, the covalent activation of phosphorylase kinase, and phosphorylase activation. Omission of external calcium eliminates phosphorylase activation without altering that of phosphorylase kinase or diminishing the increased levels of cAMP (186, 187). Thus, the activated phosphorylase kinase still requires Ca2+ for phosphorylase activation. Data with papillary muscle suggests that the level of Ca2+ required by activated phosphorylase kinase is less than that needed to stimulate contraction (183). Thus, a more stringent removal of Ca2+ is required to diminish isoproterenol-stimulated phosphorylase activation than is needed to block contraction, and isoproterenol activates phosphorylase in nonstimulated muscle without promoting contraction. These data, coupled with the observation that in the absence of P-adrenergic stimulation cardiac muscle contraction occurs without phosphorylase activation, suggest that, as with skeletal muscle, nonactivated phosphorylase kinase has a higher Ca2+ requirement than does the covalently activated enzyme. These data are in apparent conflict with in v i m +

+

+

+

+

+

+

450


results since with either purified cardiac enzyme (22), or the enzyme in a cell extract (72), an identical Ca2+ dependency has been observed for both nonactivated or activated enzyme. This is in contrast to what was observed for the skeletal-muscle enzyme where phospho-phosphorylase kinase has a higher affinity for Ca2+ (106). One important consideration of the requirement of activated phosphorylase kinase for Ca2+ during repetitive cardiac contraction comes from the data of Barovsky and Gross (188). These authors, working with isolated mouse right ventricular strips, showed that isoproterenol-stimulated phosphorylase a formation is stimulation-frequency dependent. With stimulation at the higher frequency (-3.3 Hz) isoproterenol caused phosphorylase activation which was coordinated with the expected changes in both cAMP and phosphorylase kinase. At lower frequencies (-0.2 Hz), both an increase in cAMP levels and a covalent activation of phosphorylase kinase occurred, as at the higher frequency, but phosphorylase activation was eliminated. In contrast to this, tension development was greater at the lower frequency stimulation that at the higher frequency. The higher tension development at the lower frequency indicates that total calcium during the twitch is most likely higher. The probable explanation for the absence of (measurable) phosphorylase activation at the lower stimulation frequency is that the elevated level of Ca2 is not maintained for a sufficient period of time to permit an accumulation of phosphorylase a. Thus, at the lower frequency (0.2 Hz), Ca2+ was elevated above the threshold contractile level for only 3.6 sec out of every minute, whereas at the higher frequency (3.3 Hz) the calcium was above threshold for 40 sec every minute, albeit at a lower total level than at 0.2 Hz. It is apparent that this additional dwell time for elevated Ca2+ is essential for the net (overall) formation of phosphorylase a. The response times for Ca2 -dependent phosphorylase formation are thus quite different than those for contraction. At low frequency stimulation, the elevated Ca2+ levels quite possibly result in some formation of phosphorylase a which, however, is rapidly dephosphorylated in the long relaxation time between twitches. Observable accumulation of phosphorylase a apparently occurs only if Ca2 levels are repeatedly elevated by more frequent stimulation leaving less time for dephosphorylation to occur. This difference in dwell time accounts for the different time courses of Ca2 -dependent contraction and phosphorylase a formation, despite their having very similar Ca2+ concentration dependencies. +

+

+

+

B . CORRELATION BETWEEN PHOSPHORYLASE KINASE ACTIVATION AND PHOSPHORYLATION IN INTACT TISSUES The phosphorylation of phosphorylase kinase in intact tissues was initially examined by Mayer and Krebs (141). Their study utilized skeletal muscle stimulated with epinephrine, and phosphorylation was examined following purifica-


45 1

tion of the enzyme by standard procedures. Despite being able to demonstrate phosphorylase activation and phosphorylation, the results with phosphorylase kinase were negative in that activation was observed but no epinephrine-enhanced phosphorylation detected. Several factors may have obscured the phosphorylation of phosphorylase kinase in these studies, among which was the relatively high amount of phosphate present in the unactivated enzyme and the low yield of phosphorylase kinase obtained following extensive purification. One observation of note was that the phosphate present in the unactivated phosphorylase kinase, presumably arising by exchange, achieved a higher specific activity than the y-phosphate of intracellular ATP. This suggests that phosphorylase kinase may be exposed to a different pool of ATP than that constituting the major fraction. Such an observation will need to be considered in future studies. The first study to show apparent phosphorylation of phosphorylase kinase in intact tissue was that by Yeaman and Cohen (189). For their investigation rabbit skeletal muscle was used as the tissue, animals were injected with a bolus of epinephrine, tissue was excised, the enzyme isolated by standard purification procedures, and tryptic phosphopeptides isolated. By this procedure, from epinephrine-treated animals, two phosphopeptides were identified corresponding to the major tryptic peptides phosphorylated by the CAMP-dependent protein kinase (see Section VI1,C). These data appear to demonstrate CAMPdependent protein kinase-catalyzed phosphorylation of the a-and P-subunits in the intact cell but with two reservations. A potential problem with the described experiment was that tissue was not rapidly frozen, leaving open the possibility of post homogenization events. That such activation might have occurred may be indicated by the higher values of the activity ratio obtained at pH 6.8 to that at 8.2 (189) relative to other studies (110, 111, 176, 177). Subsequent studies on the potential correlation between phosphorylase kinase activation and phosphorylation have been pursued in our laboratory with both cardiac and skeletal muscle (145, 190-192). For cardiac muscle, the Lagendorff retrograde perfusion was used. For skeletal muscle, the rat flexor digitorurn brevis was the experimental tissue (193). This latter muscle preparation is composed primarily of FOG fibers (>90%) and contains mainly the a’isozyme of phosphorylase kinase (>90%). With each experimental preparation, the tissue was rapidly frozen to permit detailed evaluation of time-course changes. The enzyme was isolated by immunoprecipitation and extensive controls were employed to insure that both the phosphorylation and the activation being measured occurred while the cells were intact [see Ref. (190) for example]. In the initial study with cardiac muscle (190), enzyme activation and total phosphorylation were examined as a function of time over a range of epinephrine concentrations, and with dephosphorylation following removal of the stimulus. With this approach a linear correlation was observed ( r = 0.94) between activation and total enzyme phosphorylation.

45 2


This study has been expanded (192) by an examination of specific subunit phosphorylation, with submaximal and near maximal stimulation by epinephrine, isoproterenol, and glucagon. With each, P-subunit phosphorylation occurs prior to a-subunit phosphorylation, but the lag between the two is dependent upon the stimulus, being greater with isoproterenol. This lag appears to be more pronounced with either the inclusion of verapamil or deletion of external Ca2+. The total level of incorporation is estimated to be -1.2-1.6 mol 32P/mol (a’P@),, compared to a basal level of 0.4-0.5 mol/mol. In both basal and maximally stimulated preparations, with most of the conditions tested, 32Pi is about equally divided between the a’-and P-subunits of phosphorylase kinase. The time course of activation appears to correlate best with that of P subunit phosphorylation, but a role for a subunit phosphorylation in activation also appears probable. In particular, following removal of the hormonal stimulus there is a brisk dephosphorylation of the P subunit but slower a subunit dephosphorylation and inactivation. With the isolated skeletal-muscle preparation (145) the phosphorylation of both the a’-and P-subunits has also been documented. Preincubation of the tissue with 32Pi-P0, leads to incorporation into both subunits, with that in the a’ subunit being 2- to 3-fold higher. Stimulation of the muscle preparation with either epinephrine or isoproterenol leads to enhanced phosphorylation of both subunits, again with a 2- to 3-fold higher level being incorporated into the a’subunit. The stoichiometry of phosphorylation into the a’-subunit is estimated to be -0.8-0.9 mol/mol (a’P$i), compared to 0.2-0.5 mol/mol in the P-subunit. No difference is observed in initial time courses of phosphorylation between the two subunits, although that of the P-subunit plateaued earlier. The best correlation with activation appears to be with total phosphorylation (i.e., a’ P) (I = 0.82-0.97). The data that are being accumulated for in vivo phosphorylation and activation of phosphorylase kinase in general appears to be supportive of what has been learned from in vitro studies, but with some differences. In both, the CAMPdependent phosphorylation of both a- and P-subunits appears clearly to be a prime feature of control. If anything, the higher level of skeletal-muscle a’subunit phosphorylation in the intact muscle preparation points to a greater role for that subunit, rather than the 9-subunit, in control. One common feature of both cardiac- and skeletal-muscle intact tissue studies is that the level of subunit phosphorylation appears substantially less than stoichiometric, despite high levels of hormonal stimulation. This observation is consistent with the questions raised concerning the potential importance of the initial phosphorylation events in producing some of the major changes in enzyme conformation and perhaps thus regulation. Clearly, however, many studies remain to be done before a full understanding of the regulation and roles of phosphorylase kinase in muscle function can be appreciated.

+

453


ACKNOWLEDGMENTS This chapter is dedicated to Dr. George E. Drummond and Dr. Steven E. Mayer, without whose contributions our knowledge of phosphorylase kinase would be much less. This work was supported by AM 13613and AM 21019. The authors greatly appreciate the in-depth comments of K.Angelos, L. Anderson, H. C. Cheng, R. Cooper, L. Garetto, P. Hallenbeck, and C. Ramachandran.

REFERENCES 1. Sutherland, E. W. (1950). Recent Prog. Horm. Res. 5, 441-463.

2. 3. 4. 5. 6.

7. 8. 9. 10.

11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

Cori, C. F., and Cori, G. T. (1945).JBC 158,341-345. Burnett, G., and Kennedy, E. P. (1954). JBC 211, 969-980. Krebs, E. G., and Fischer, E. H. (1956). BBA 20, 150-157. Lipkin, D., Markham, It., and Cook, W. H. (1959). J . Am. Chem. SOC. 81, 6075. Robison, G. A,, Butcher, R. W., and Sutherland, E. W. (1971). “Cyclic AMP.” Academic Press, New York. Krebs, E. G., Love, D. S., Bratvold, G. E., Trayser, K. A., Meyer, W. L., and Fischer, E. H. (1964). Biochemistry 3, 1022-1033. DeLange, R. J . , Kemp, R. G., Riley, W. D., Cooper, R. A,, and Krebs, E. G. (1968). JBC 243, 2200-2208. Walsh, D. A,, Perkins, J. P., and Krebs, E. G. (1968). JBC 243, 3763-3774. Chan, K. J., and Graves, D. J. (1984). In “Calcium and Cell Function” (W. Y. Cheung, ed.), Vol. 5 , pp. 1-31. Academic Press, New York. Carlson, G. M., Bechtel, P. I., and Graves, D. J. (1979). Adv. Enzymol. Relat. Areas Mol. Biol. 50, 41-1 15. Malencik, D. A,, and Fischer, E. H. (1982). In “Calcium and Cell Function” (W. Y. Cheung, ed.), Vol. 2, pp. 161-188. Academic Press, New York. Hayakawa, T., Perkins, J. P., Walsh, D. A., and Krebs, E. G. (1973). Biochemistry 12,567573. Cohen, P. (1973). EJB 34, 1-14. Sharma, R. K., Tam, S. W., Waisman, D. M., and Wang, J. H. (1980). JBC 255, 1110211 105. Jennissen, H. P., Horl, W. H., and Heilmeyer, L. M. G., Jr. (1973). Hoppe-Seyler’s Z. Physiol. Chem. 354, 236-237. Jennissen, H. P., and Heilmeyer, L. M. G., Jr. (1975). Biochemistry 14, 754-760. Hallenbeck, P. C., and Walsh, D. A. (1983). JBC 258, 13493-13501. Gross, S. R., and Brornwell, K. (1977). ABB 184, 1-1 1. Pocinwong, S., Blum, H., Malencik, D., and Fischer, E. H. (1981). Biochemistry 20, 72197226. Tam, S. W., Sharma, R. K., and Wang, J. H. (1982). JBC 257, 14907-14913. Cooper, R. H., Sul, H. S.,McCullough, T. E., and Walsh, D. A . (1980). JBC 255, 1179411801.

22a. Nikolaropoulos, S., and Sotiroudis, T. G. (1985). EJB 151, 467-473. 23. Chrisman, T. D., Jordan, J. E., and Exton, J. H. (1982). JBC 257, 10798-10804. 24. Pohlig, G., Wingender-Drissen, R., and Becker, J. U. (1983). BBRC 114, 331-338. 25. Wingender-Drissen, R., and Becker, J. U. (1983). FEBS Lett. 163, 33-36. 26. Jennissen, H. P., and Heilmeyer, L. M. G., Jr. (1974). FEBS Lett. 42, 77-80. 27. Crabb, J . W., and Heilmeyer, L. M. G., Jr. (1984). JBC 259, 6346-6350, 14314.

454


28. Reimann, E. M., Titani, K., Ericsson, L. H., Wade, R. D., Fischer, E. H., and Walsh, K. A. (1984). Biochemistry 23, 4185-4192. 29. Grand, R. J., Shenolikar, S . , and Cohen, P. (1981). EJB 113, 359-367. 30. Burchell, A,, Cohen, P. T. W., and Cohen, P. (1976). FEES Lett 67, 17-22. 30a. Pickett-Gies, C. A., Carlsen, R., Angelos, K. L., and Walsh, D. A. (1986). JBC (in press). 30b. Lawrence, J. C., Jr., Krsek, J. A., Salsgiver, W. J., Hiken, J. F., Salmons, S . , and Smith, R. L. (1986). Am. J. Physiol. 250, C84-C89. 31. Schwartz, A., Entman, M. L., Kaniike, K., Lane, L. K., Van Winkle, B., and Bornet, E. P. (1976). BBA 426, 57-72. 32. Hod, W. H., and Heilmeyer, L. M. G., Jr. (1978). Biochemistry 17, 766-772. 33. Le Peuch, C. J., Le Peuch, D. A., and Demaille,G. (1982). Ann. N.Y. Acud. Sci. 402, 549557. 34. Hessova, Z., Varsanyi, M.,and Heilmeyer, L. M. G.,Jr. (1985). EJB 146, 107-1 15. 35. Cohen, P. (1978). Curr. Top. Cell. Re& 14, 117-196. 35a. Trempe, M. R., Carlson, G. M., Hainfeld, J. F., Furcinitti, P. S., and Wall, J. S . (1986). JBC 261, 2882-2889. 36. Lambooy, P. K., and Steiner, R. F. (1982). ABB 213, 551-556. 37. Picton, C., Klee, C. B., and Cohen, P. (1980). EJB 111, 553-561. 38. Fitzgerald, T. J., and Carlson, G.M. (1984). JBC 259, 3266-3274. 39. Picton, C., Klee, C. B., and Cohen, P. (1981). Cell Calcium 2, 281-294. 40. Chan, K.-F. J., and Graves, D. J. (1982). JBC 257, 5939-5947. 41. Skuster, J. R., Chan, K.-F. J., and Graves, D. J. (1980). JBC 255, 2203-2210. 42. Kee, S . M., andGraves, D. J. (1985). FP44, 1074. 43. Cohen, P., Burchell, A., Foulkes, J. G.,Cohen, P. T. W., Vanaman, T. C., and Nairn, A. C. (1978). FEES Leu. 92, 287-293. 43a. Jennissen, H. P., Horl, W. H., Groschel-Stewart, U.,Velick, S. F., and Heilmeyer. L. M. G., Jr. (1976). I n “Metabolic Interconversions of Enzymes” (S. Shaltiel, ed.), pp. 19-26. Springer-Verlag, Berlin and New York. 44. Hayakawa, T., Perkins, J. P., and Krebs, E. 0. (1973). Biochemistry 12, 574-580. 45. Chan, K.-F. J., and Graves, D. J. (1982). JBC 257, 5948-5955. 46. Chan, K.-F. J., and Graves, D. J. (1982). JBC 257, 5956-5961. 47. Gulyaeva, N. B., Vul’fson, P. L., and Severin, E. S . (1977). Biokhimiyu 43, 373-381. 48. King, M. M., Carlson, G.M., and Haley, B. E. (1982). JBC 257, 14058-14065. 48a. King, M. M., and Carlson, G.M. (1981). Biochemistry 20, 4382-4387. 48b. King, M. M., and Carlson, G.M . (1982). FEES Lett. 140, 131-134. 49. Fischer, E. H., Alaba, J. O., Brautigan, D. L., Kerrick, W. G. D., Malencik, D. A,, Moeschler, H. J., Picton, C., and Pocinwong, S . (1978). In “Versatility of Proteins” (C. H. Li, ed.), pp. 133-145. Academic Press, New York. 50. Killilea, S . D., and Ky, N. M. (1983). ABB 221, 333-342. 50a. Sotiroudis, T. G.,and Nikolaropoulos, S . (1984). FEES Leu. 176, 421-425. 50b. Crabb, J. W., Sotiroudis, T. G.,and Heilmeyer, L. M. G.,Jr. (1986). JBC (submitted for publication). 51. Carlson, G.M., and Graves, D. J. (1976). JBC 251, 7480-7486. 52. Dickneite, G.,Jennissen, H. P., and Heilmeyer, L. M. G., Jr. (1978). FEES Lett. 87, 297302. 53. King, M. M., and Carlson, G.M. (1981). Biochemistry 20, 4387-4393. 54. Kilimann, M. W., and Heilmeyer, L. M. G.,Jr. (1982). Biochemistry21, 1727-1734. 55. Kilimann, M. W., and Heilmeyer, L. M. G.,Jr. (1982). Biochemistry 21, 1735-1739. 55a. Georgoussi, Z., and Heilmeyer, L. M. G.,Jr. (1986). Biochemistry (in press).


455

55b. Hessova, Z., Thieleczek, R., Varsanyi, M., Falkenberg, F. W., and Heilmeyer, L. M. G., Jr. (1985). JBC 260, 10111-101 17. 56. Grand, R. J., Shenolikar, S., and Cohen, P. (1981). EJB 113, 359-367. 57. Heilmeyer, L. M. G., Jr., Groschel-Stewart, U., Jahnke, U.,Kilimann, M. W., Kohse, K. P., and Varsanyi, M. (1980). Adv. Enzyme Regul. 18, 121-144. 58. Kilimann, M., and Heilmeyer, L. M. G., Jr. (1977). EJB 73, 191-197. 59. Kohse, K. P., and Heilmeyer, L. M. G., Jr. (1981). EJB 117, 507-513. 60. Stull, J. T., Nunnally, M. H., and Michnoff, C. H. (1986). This volume, Chapter 4. 61. Wang, J. H., Stull, J. T., Huang, T.-S., and Krebs, E. G. (1976). JBC 251, 4521-4527, 62. Carlson, G. M., and Graves, D. J . (1976). JBC 251, 7480-7486. 63. Graves, D. J., Hayakawa, T., Horvitz, R. A,, Beckman, E., and Krebs, E. G. (1973). Biochemisrry 12, 580-585. 64. Sul, H. S., Dirden, B., Angelos, K. L., Hallenbeck, P., and Walsh, D. A. (1983). “Methods in Enzymology,” Vol. 99, pp. 250-259. 64a. Jennissen, H. P., Petersen-Von Gehr, J. K. H., and Botzet, G. (1985). EJB 147, 619-630. 65. Clerch, L. B., and Huijing, F. (1972). BBA 268, 654-662. 66. Kim, G., and Graves, D. J. (1973). Biochemisrry 12, 2090-2095. 67. Singh, T. J., Akatsuka, A., and Huang, K.-P. (1982). ABB 218, 360-368. 68. Villa-Palasi, C., and Wei, S . H. (1970). PNAS 67, 345-350. 69. Tabatabai, L. B., and Graves, D. J. (1978). JBC 253, 2196-2202. 70. Meyer, W. L., Fischer, E. H., and Krebs, E. G. (1964). Biochemistry 3, 1033-1039. 71. Cheng, A , , Fitzgerald, T. J., and Carlson. G. M. (1985).JBC 260, 2535-2542. 72. Hayes, J. S., and Mayer, S. E. (1981). Am. J. Physiol. 240, E340-E349. 73. Ramachandran, C., Angelos, K. L., and Walsh, D. A. (1982). JBC 257, 1448-1457. 74. King, M. M., and Carlson, G. M. (1981). ABB 209, 517-523. 75. King, M. M., and Carlson, G. M. (1981). JBC 256, 11058-11064. 75a. Hallenbeck, P.. and Walsh, D. A. (1986).JBC261,5442-5449. 76. Stull, J. T., England, P. J., Brostrom, C. 0.. and Krebs, E. G. (1972). Cold Spring Harbor Symp. Quant. Biol. 37, 263-265. 77. Tessmar, G., and Graves, D. J. (1973). BBRC 50, 1-7. 78. Krebs, E. G., Kent, A. B., and Fischer, E. H. (1958). JBC 231, 73-83. 79. Shizuta, Y., Khandelwal, R. L., Maller, J. L., Vandenheede, J. R., and Krebs, E. G. (1977). JBC 252, 3408-3413. 80. Nolan, C., Novoa, W. B., Krebs, E. 0 . . and Fischer, E. H. (1964).Biochemistry 3,542-551. 81. Tessmar, G. W., Skuster, J. R., Tabatabai, L. B., and Graves, D. I. (1977).JBC 252, 56665671. 82. Graves, D. J. (1983). “Methods in Enzymology,” Vol. 99, pp. 268-278. 83. Viriya, J., and Graves, D. I . (1979). BBRC 87, 17-24. 84. Chan, K.-F. J., Hurst, M. V., and Graves, D. J. (1982). JBC 257, 3655-3659. 85. Kemp, B. E., and John, M. J. (1981). Cold Spring Harbor Conf. Cell Proliferarion 8, 331342. 86. Soderling, T. R., Srivastava, A. K., Bass, M. A., and Khatra, B. (1979). !“AS 76, 25362540. 87. Walsh, D. A,, Perkins, J. P., Brostrom, C. 0..Ho, E. S., and Krebs, E. G. (1971).JBC 246, 1968- 1976. 88. King, M. M., Fitzgerald, T. J., and Carlson, G. M. (1983). JBC 258, 9925-9930. 89. DePaoli-Roach, A. A,, Roach, P. J., and Lamer, J. (1979). JBC 254, 4212-4219. 90. Walsh, K. X., Millikin, D. M., Schlender, K. K., and Reimann, E. M. (1979). JBC 254, 661 1-6616.

456


90a. Akatsuka, A., Singh, T. J., and Huang, K.-P. (1984). JBC 259, 7878-7883. 91. Stull, J. T., Brostrom, C. O., and Krebs, E. G. (1972). JBC 247, 5272-5274. 92. Perry, S. V., and Cole, H. A. (1974). BJ 141, 733-743. 93. Moir, A. J. G., Cole, H. A., and Perry, S . V. (1977). BJ 161, 371-382. 94. St. Louis, P. J., and Sulakhe, P. V. (1977). Eur. J. Pharmacol. 43, 277-280. 94a. De Paoli-Roach, A. A., Bingham, E. W., and Roach, P. J. (1981). ABB 212, 229-236. 94b. Singh, T. J., Akatsuka, A., and Huang, K.-P. (1983). FEES Lett. 159, 217-220. 95. Soderling, T. R., Sheorain, V. S . , and Ericsson, L. H. (1979). FEES Leu. 106, 181-184. 96a. Angelos, K. L., and Walsh, D. A. (unpublished observation). 97. Meyer, F., Heilmeyer, L. M. G., Jr., Haschke, R. H., and Fischer, E. H. (1970). JBC 245, 6642-6648. 98. Jennissen, H. P., Horl, W. H., Groschel-Stewart, U., Velick, S . F., and Heilmeyer, L. M. G., Jr. (1975). In “Metabolic Interconversion of Enzymes” (S. Shaltiel, ed.), pp. 19-26. Springer-Verlag, Berlin and New York. 99. Dombradi, V. K., Silberman, S . R., Lee, E. Y. C., Caswell, A. H., and Brandt, N. R. (1984). ABB 230, 615-630. 100. Flockhart, D. A., Freist, W., Hoppe, J., Lincoln, T. M., and Corbin, J. D. (1984). EJB 140, 289-295. 100a. Livanova, N. B., Silonova, G. V., Solovyeva, N. V., Andreeva, 1. E., Ostrovskaya, M. V., and Poglazov, B. F. (1983). Biochem. Int. 7, 95-105. 101. Huang, T. S . , Bylund, D. B., Stull, J. T., and Krebs, E. G. (1974). FEES Len. 42,249-252. 102. Heilmeyer, L. M. G., Jr., Meyer, F., Haschke, R. H., and Fischer, E. H. (1970). JBC 245, 6649-6656. 103. Gergely, P., Vereb, G., and Bot, G. (1975). Arch. Biochem. Biophys. Acad. Sci. Hung. 10, 153. 104. Burger, D., Cox, J. A., Fischer, E. H., and Stein, E. A. (1982). BBRC 105, 632-637. 105. Burger, D., Stein, E. A., and Cox, J. A. (1983). JBC 258, 14733-14739. 106. Cohen, P. (1980). EJB 111, 563-574. 107. Brostrom, C. O., Hunkeler, F. L., and Krebs, E. G. (1971). JBC 246, 1961-1967. 108. Ozawa, E., Hosoi, K., and Ebashi, S . (1967). J . Biochem. (Tokyo)61, 531-533. 109. Ebashi, S., Makoto, E., and Ohtsuki, I. (1969). Q. Rev. Biophys. 2, 351-384. 110. Drummond, G. E., Hanvood, J. P., and Powell, C. A. (1969). JBC 244, 4235-4240. 111. Stull, J. T., and Mayer, S. E. (1971). JBC 246, 5716-5723. 112. Vandenheede, J. R., De Wulf, H., and Merlevede, W. (1979). EJB 101, 51-58. 113. Keppens, S., Vandenheede, J. R., and De Wulf, H. (1977). BBA 496,448-457. 114. Chenington, A. D., Hundley, R. F., Dolgin, S . , and Exton, J. H. (1977). J . Cyclic Nucleotide Res. 3, 263-27 1. 115. Hutson, N. J., Brumley, F. T., Assimacopoulos, F. D., Harper, S . C., and Exton, J. H. (1976). JBC 251, 5200-5208. 116. Cohen, S . M., and Burt, C. T. (1977). PNAS 74, 4271-4275. 117. Heilmeyer, L. M. G., Jr., Jahnke, U., Kilimann, M. W., Kohse, K. P., and Sperling, J. E. (1981). Cold Spring Harbor Conf. Cell Proliferation 8, 321-329. 118. DePaoli-Roach, A. A., Gibbs, J. B., and Roach, P. J. (1979). FEES Lett. 105, 321-324. 119. Shenolikar, S., Cohen, P. T. W., Cohen, P., Naim, A. C., and Perry, S. V. (1979). EJB 100, 329-337. 120. Wilkinson, J. (1980). EJB 103, 179-188. 121. Potter, J. D., Johnson, J. D., Dedman, J. R., Schreiber, W. E., Mandel, F., Jackson, R. L., and Means, A. R. (1977). I n “Calcium Binding Proteins and Calcium Functions” (R. H. Wasseman, R. A. Corradino, E. Carafoli, R. H. Kretsinger, D. H. MacLennan, and F. L. Siegel, eds.), pp. 239-250. Elsevier/North-Holland, New York.


457

Leavis, P. C., and Kraft, E. L. (1978). ABB 186, 411-415. Sigel, P., and Pette, D. (1969). . I Hisrochem. . Cyrochem. 17, 225-237. Pette, D. (1975). Acru Hisrochem. Suppl. 14, 47-68. Yoshikawa, K., Usui, H., Imam, M., Takeda, M., and Ebashi, S. (1983). EJB 136,413-419. Wanson, J.-C., and Drochmans, P. (1972). J . Cell Biol. 54, 206-224. DiMauro, S., Trojaborg, W., Gambetti, P., and Rowland, L. P. (1971). ABB 144,413-422. Bergamini, C., Buc, H., and Morange, M. (1977). FEES Lerr. 81, 166-172. 128a. Reddy, N. B., Oliver, K. L., Festoff, B. W., and Engel, W. K. (1978). BBA 540, 371-378. 129. Kasvinsky, P. J., Madsen, N. B., Sygusch, J., and Fletterick, R. J. (1978). JBC 253, 33433351. 130. Madsen, N. B., Kasvinsky, P. J., and Fletterick, R. J. (1978). JBC 253, 9097-9101. 131. Steiner, R. F., and Marshall, L. (1982). BBA 707, 38-45. 132. Hallenbeck, P., and Walsh, D. A., unpublished observation. 133. Preiss, J., and Walsh, D. A. (1981). In “Biology of Carbohydrates” (V. Ginsberg, ed.), Vol. 1, pp. 199-314. Wiley, New York. 134. Busby, S. J. W., Gadian, D. G., Griffths. J. R., Radda, G.K., and Richards, R. E. (1976). EJB 63, 23-3 1. 135. Srivastava, A. K., Khatra, B. S., and Soderling, T. R. (1980). ABB 205, 291-296. 136. Gergely, P., Vereb, G.,and Bot, G.(1976). BBA 429, 809-816. 137. Fischer, E. H., and Krebs, E. G.(1966). FP 25, 1511- 1520. 138. Entman, M. L., Kaniike, K., Goldstein, M. A., Nelson, T. E., Bornet, E. P., Futch, T. W., and Schwartz, A. (1976). JBC 251, 3140-3146. 139. Entman, M. L., Bornet, E. P., Van Winkle, W. B., Goldstein, M. A,, and Schwartz, A. (1977). J. Mol. Cell. Cardiol. 9, 515-528. 140. Kayikawa, N., Kishimoto, A., Shiota, M., and Nishizuka, Y. (1983). “Methods in Enzymology,” Vol. 102, pp. 279-290. 141. Mayer, S. E., and Krebs, E. G. (1970). JBC 245, 3153-3160. 142. Pickett-Gies, C. A,, and Walsh, D. A. (1985). JBC 260, 2046-2056. 143. Kilimann, M. W., Schnackerz, K. D., and Heilmeyer, L. M.G.,Jr. (1984). Biochemisrry 23, 112-117. 144. Cooper, R. H., Sul, H. S., and Walsh, D. A . (1981). JBC 256, 8030-8038. 145. Pickett-Gies, C. A., Carlsen, R., Anderson, L. J., Angelos, K. L., and Walsh, D. A. (1986). JBC 261, (in press). 146. Lincoln, T. M., and Corbin, J. D. (1977). PNAS 74, 3239-3243. 147. Cohen, P. (1980). FEBS Lerr. 119, 301-306. 148. Waisman, D. M., Singh, T. J., and Wang, J. H. (1978). JBC 253, 3387-3390. 149. Singh, T. J., Akatsuka, A., and Huang, K.-P. (1982). JBC 257, 13379-13384. 149a. Singh, T. J., Akatsuka, A., and Huang, K. P. (1984). JBC 259, 12857-12864. 150. Kishimoto, A,, Takai, Y., and Nishizuka, Y. (1977). JBC 252, 7449-7452. 151. Takai, Y., Kishimoto, A., Iwasa, Y., Kawahara, Y., Mori, T., andNishizuka, Y. (1979).JBC 254, 3692-3695. 152. Hallenbeck, P., Ramachandran, C., and Walsh, D. A., unpublished. 153. Cook, P. F., Neville, M. E., Vrana, K. E., Hartl, F. T., and Roskoski, R. R. (1982). Biochemistry 21, 5794-5799. 154. Singh, T. J., and Wang, J. H. (1977). JBC 252, 625-632. 155. Cohen, P., Watson, D. C., and Dixon, G.H. (1975). EJB 51, 79-92. 156. Sul, H. S., Cooper, R. H., Whitehouse, S., and Walsh, D. A. (1982). JBC257,3484-3490. 157. Cox, D. E., and Edstrom, R. D. (1982). JBC 257, 12728-12733. 158. Singh, T. J., and Huang, K.-P. (1985). BBRC 130, 1308-1313. 159. Yeaman, S. J., Cohen, P., Watson, D. C., and Dixon, G. H. (1977). BJ 162, 411-421. 122. 123. 124. 125. 126. 127. 128.

458


159a. Cheng, H.-C., Kemp, B. E., Pearson, R. B., Smith, A. J., Misconi, L., Van Patten, S. M., and Walsh, D. A. (1986). JBC 261, 989-992. 160. Sul, H. S., and Walsh, D. A. (1982). JBC 257, 10324-10328. 161. Ganapathi, M. K . , and Lee, E. Y. C. (1984). ABB 233, 19-31. 161a. Ramachandran, C., Pickett-Gies, C. A., Goris, J., Waelkens, E., Merlevede, W., and Walsh, D. A. (1985). Adv. Protein Phosphatases 2, 355-374. 161b. Ramachandran, C., Goris, J., Waelkens, E., Merlevede, W., and Walsh, D. A. (1986). JBC 261, (in press). 162. Cohen, P., and Antoniw, J. F. (1973). FEES Lett. 34, 43-47. 163. Stewart, A. A., Hemmings, B. A,, Cohen, P., Goris, J., and Merlevede, W. (1981). EJB 115,

197-205. 164. Ganapathi, M. K., Silberman, S. R., Paris, H., and Lee, E. Y. C. (1981). JBC 256, 32133217. 165. Jurgensen, S . , Shacter, E., Huang, C. Y., Chock, P. B., Yang, S. D., Vandenheede, I. R., and Merlevede, W. (1984). JBC 259, 5864-5870. 166. Vandenheede, J. R., Yang, S. D., and Merlevede, W. (1981). JBC 256, 5894-5900. 167. Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 255-261. 168. Ingebritsen, T. S. Foulkes, Z. G., and Cohen, P. (1983). EJB 132, 263-274. 169. Ingebritsen, T. S . , Blair, J., Guy, P., Witters, L., and Hardie, D. C. (1983). EJB 132, 275281. 170. Pato, M. D., Adelstein, R. S., Crouch, D., Safer, B., Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 283-287. 171. Stewart, A,, Ingebritsen, T. S., and Cohen, P. (1983). EJB 132, 289-295. 172. Pelech, S., Cohen, P., Fisher, M. Z., Pogson, C., El-Maghrabi, R., and Pilkus, S. J. (1984). EJB 145, 39-45. 173. Ingebritsen, T. S . , Stewart, A. A.. and Cohen, P. (1983). EJB 132, 297-307. 173a. DiSalvo, J., and Merlevede, W. (1985). I n “Advances in Protein Phosphatases,” Vol. 1. Leuven Press. 173b. DiSalvo, J., and Merlevede, W. (1985). I n “Advances in Protein Phosphatases,” Vol. 2. Leuven Press. 173c. Tsuchiya, M., Tanigawa, Y., Ushiroyama, T., Matsuura, R., and Shimoyama, M. (1985). EJB 147, 33-40. 174. Williamson, J. R., Cooper, R. H., and Hoek, I. B. (1981). BBA 639, 243-295. 175. Assimacopoulos-Jeannet, F. D., Blackmore, P. F., and Exton, J. H. (1977). JBC 252,26622669. 176. Gross, S. R., and Johnson, R. M. (1980). J. Pharmcol. Exp. Ther. 214, 37-44. 177. Posner, J. B., Stem, R. S . , and Krebs, E. G. (1985). JBC 240, 982-985. 178. Danforth, W. H., and Helmreich, E. (1964). JBC 239, 3133-3138. 179. Danforth, W. H., and Lyon, J. B. (1964). JBC 239, 4047-4050. 180. Friesen, A. J., Allen, G., and Valadares, J. R. (1967). Science 155, 1108-1 109. 181. Friesen, A. J., Oliver, N., and Allen, G. (1969). Am. J . Physiof. 217,445-450. 182. Dobson, J. G., and Mayer, S. E. (1973). Circ. Res. 33, 412-420. 183. Dobson, J. G., Ross, J., and Mayer, S. E. (1976). Circ. Res. 39, 388-395. 184. Hammermeister, K. E., Yunis, A,, and Krebs, E. G. (1965). JBC 240, 986-991. 185. Namm, D. H., and Mayer, S. E. (1968). Mol. Pharmacol. 4, 61-69. 186. Namm, D. H., Mayer, S. E., and Maltbie, M. (1968). Mol. Pharmcol. 4, 522-530. 187. Mayer, S. E., Namm, D. H., and Rice, L. (1970). Circ. Res. 26, 225-233. 188. Barovsky, K . , and Gross, S. R. (1981). J . Pharmacol. Exp. Ther. 217, 326-332. 189. Yeaman, S. J . , and Cohen, P. (1975). EJB 51, 93-104.


459

190. McCullough, T. E., and Walsh, D. A. (1979). JBC 254, 7345-7352. 191. Sul, H. S . , Cooper, R. H., McCullough, T. E., Pickett-Gies, C. A , , Angelos, K. L., and Walsh, D. A . (1981). Cold Spring Harbor Conf. Cell Proliferation 8, 343-355. 192. Angelos, K . L., Ramachandran, C., and Walsh, D. A. (1986). JBC (in press.) 193. Carlsen, R. C., Larson, D. B., and Walsh, D. A. (1985). Can. J . Pharmacol. Physiol. 63, 958-965.


Muscle Glycogen Synthase PHILIP COHEN Department of Biochemistry The University Dundee DDI 4HN. United Kingdom

1. Introduction .................................................... 11. Structure of Glycogen Synthase from Mammalian Skeletal Muscle . . . . . . . . 111. Glycogen Synthase Kinases in Mammalian Skeletal Muscle . . . . . . . . . . . . . A. Cyclic AMP-Dependent Protein Kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Phosphorylase Kinase . . . . .

D. Glycogen Synthase Kinase-3 ............................. E. Glycogen Synthase Kinase-4 F. Glycogen Synthase Kinase-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Casein Kinase-I ....................... I. Diacylglycerol-DependentProtein Kinase . . . . . . . . . . . . . . . . . . . . . . . . . IV. Effect of Phosphorylation on the Activity of Skeletal-Muscle Glycogen Synthase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Synergism between Glycogen Synthase Kinase-3 and Glycogen Synthase Kinase-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V1. The Glycogen Synthase Phosphatases in Skeletal Muscle . . . . . . . . . . . . . . . A. Structure of Protein Phosphatase-1 and Protein Phosphatase-2A . . . . . . . B. Regulation of Protein Phosphatase-1 and Protein Phosphatase-2A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Phosphorylation State of Skeletal-Muscle Glycogen Synthase in Vivo . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Effect of Epinephri Synthase in Vivo . B. Effect of Insulin on Synthase in Vivo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... C. Distribution of Phosphate within the Peptide Chain .

462 462 464 464 464 465 466 467 467 468 469 469 469 47 1 472 473 474 478 478 479 48 1

46 I THE ENZYMES. Vol. XVIl Copyright 0 1986 by Academic Press. Inc. All rights of repmduclion in any lomi reserved.

462

PHILIP COHEN

VIII. Interpretation of in Vivo Phosphorylation Experiments . . . . . . . . . . . . . . . . . . A. Resting Muscle in the Absence of Epinephrine B. Resting Muscle in the Presence of Epinephrine .................... C. The Effect of Insulin ..... .................. D. Muscle Contraction ........................... . . . . . . . . E. Regulation by Glycogen . . . . . . . . . . . ................... References ........................

1.

486 487 488 489 492 493 493

Introduction

Glycogen synthase was the third enzyme shown to be regulated by a phosphorylation-dephosphorylation mechanism. Following the discovery that glycogen phosphorylase ( 1 ) and phosphorylase kinase (2) were activated by phosphorylation, Lamer and co-workers found that glycogen synthase could exist in two forms in mammalian skeletal muscle. One possessed little activity without glucose 6-phosphate (G6P), whereas the other was almost fully active in the absence of this allosteric activator (3).The conversion of glycogen synthase from a G6P-independent to a G6P-dependent form was shown to require MgATP and a further protein, and to be stimulated by cyclic adenosine 3’5’-monophosphate (CAMP) (3-5). The basis for these effects became clearer following the purification of glycogen synthase to homogeneity, when the enzyme was shown to be phosphorylated by CAMP-dependent protein kinase (6, 7). However, in 1971 Smith er al. (8) reported that the G6P-dependent form contained -6 mol phosphate/86 kDa subunit, and with the subsequent identification of additional glycogen synthase kinases (9-13) it became clear that glycogen synthase was regulated by multisire phosphorylation. These discoveries provided a major stimulus for research over the subsequent decade. Progress since 1982 has been particularly marked and is reviewed in this chapter.

II. Structure of Glycogen Synthase from Mammalian Skeletal Muscle

The glycogen synthase subunit migrates on SDS-polyacrylamide gel electrophoresis (SDS-PAGE) with an apparent molecular mass of -86 kDa [reviewed in Ref. ( 1 2 ) ] . The primary structure of the first 29 (13) and last 124 residues (14) of the rabbit skeletal-muscle enzyme have been determined (Fig. l ) , establishing that the subunits are identical. The smallest active species is a tetramer [reviewed in Ref. (12)], although the tetrameric species has a strong tendency to aggregate. As discussed in Section 11, A, all the phosphorylation sites that have been sequenced are contained within the N- and C-terminal regions. Residues in the N-terminal cyanogen bromide peptide (CB-N) are there-

I I.

463

MUSCLE GLYCOGEN SYNTHASE T

C

CB-N

PLSRTLSVSSLPGLEDYEDEFDLENSVLF

'

e l le

CBX

A

LAKAFP0

e

38

20

i:Y E P H E A D A20T Q G Yi' 'R

I im

T

Y

I e

93

1 PR PA

II

e

70 iP V P P fP P 5 L fP R Hp S Pf P H Q fP E 0 E E5lE P R 0 G L P E E 60 D G E R Y D EDE E A A K 110

120

i

ORRNIRAPQYPRRASCTSSSGGSKRSNSVDTSSLSTPSEPLSSAPSLGEERN

FIG. 1. Primary structures of the N-terminal and C-terminal of the cyanogen bromide peptides (CB-N and CB-C) of rabbit skeletal-muscle glycogen synthase. The sequence of residues 108-124 in CB-C are unpublished results from this laboratory, and differ from the sequence published in (1.5) at a number of positions. Residue 124 appears to be the C-terminus of glycogen synthase. Sites phosphorylated in vitro are denoted by P, and the positions at which the native enzyme is cleaved by trypsin and chymotrypsin by T and C, respectively. Data are from Refs. (/3-/.5).

fore prefixed by N (Nl, N2, N3, etc.) and those in the C-terminal cyanogen bromide fragment (CB-C) by C (Cl, C2, C3, etc.). Glycogen synthase purified in this laboratory contains a second protein component of apparent molecular mass 44 kDa (16). The molar ratio of the 86-kDato the 44-kDa-species is approximately 2 : 1 . Whether the 44-kDa protein interacts with glycogen synthase or is merely an impurity is unresolved. The 44kDa component is not phosphorylated by any glycogen synthase kinase. The N- and C-terminal regions that have been sequenced contain all the phosphorylation sites (Section 111) and are also extremely sensitive to proteinases. Brief incubation of the native enzyme with low concentrations of trypsin initially cleaves the peptide bonds N4-N5, C75-C76, C78-C79, and C84C85, followed by C39-C40 (14, 17). Cleavage in the region C75-C85 is accompanied by a decrease in apparent molecular mass from 86 kDa to 77 kDa, and cleavage at C39-C40 by a further reduction to 69 kDa (14), as judged by SDSPAGE. Brief incubation with chymotrypsin cleaves the peptide bond C23-C24 specifically, reducing the apparent molecular mass from 86 kDa to 67 kDa (14). However the actual molecular mass of the peptide C24-C124 is only 11 kDa (Fig. 1). This peptide is extremely hydrophilic and behaves in solution as a random coil (18). Furthermore, it contains only two lysyl residues and is unlikely to bind SDS as well as normal globular proteins. These observations indicate that the unusual C-terminal region of glycogen synthase reduces its mobility on SDSPAGE. Consequently the subunit may not be as large as 86 kDa. Brief incubation of glycogen synthase with very low concentrations of subtilisin initially cleaves the peptide bond N6-N7, enhancing activity in the presence of G6P about 5-fold (17). Trypsin also causes a transient (1 .Sfold) rise in activity in the presence of G6P (17, 19) presumably resulting from cleavage of N4-N5 (17). In contrast, cleavage at the C-terminal region is accompanied by inactivation, especially in the absence of G6P (19-22).

464

PHILIP COHEN

111. Glycogen Synthase Kinases in Mammalian Skeletal Muscle

It is clear that skeletal-muscle glycogen synthase can be phosphorylated in virro by at least 10 protein kinases. Indeed, of protein kinases that have been tested, only myosin light chain kinase is unable to phosphorylate the enzyme (23, 24). In this section, the protein kinases that act on glycogen synthase and the sites that they phosphorylate are identified. A number of these protein kinases are described elsewhere in this volume, and readers are referred to other chapters for more detailed accounts of their structure and properties. A. CYCLICAMP-DEPENDENT PROTEINKINASE The phosphorylation of glycogen synthase by CAMP-dependentprotein kinase reaches different plateau values, depending on the concentration of protein kinase in the incubation (12, 25). Up to an incorporation of =2 mol/subunit nearly all the phosphate is incorporated into the seryl residues N7 (site-2), C87 (site-la) and ClOO (site-lb). The initial rate of phosphorylation of site-la is 7- to 10-fold faster than site-2 and 15- to 20-fold faster than site-lb (26). Sheorain et al. (27) have reported that additional sites can be phosphorylated when extremely high concentrations of CAMP-dependent protein kinase (2.5 pl4) are employed. Under these forcing conditons, the total amount of phosphate incorporated exceeded 3 mol/subunit, and following exhaustive digestion with trypsin, 32P-labeled peptides were resolved by reverse-phase high-performance liquid chromatography (HPLC). These experimentsrevealed the presence of 32Pradioactivity eluting at the positions expected for the tryptic peptides C25-C39 and C40-C53. The maximum levels of phosphorylation of these two regions were estimated to be 0.6 mol/subunit (C25-C39) and 0.2 mol/subunit (C40C53). Automated Edman degradation of C40-C53 showed a “burst” of 32Pradioactivity at the third cycle, suggesting that C42 was phosphorylated. The location of the phosphate in region C25-C39 was not determined. Although these results may well be correct, primary structure analysis is needed to substantiate the conclusions because the peptides were not obtained in pure form. For example phosphorylation of C42 may have rendered the peptide bond C39-C40 resistant to trypsin (see Section 111, G). Consequently, the peptide assigned at C25-C39 could have been C25-C53. If this were true, phosphorylation of C43 might be much more extensive and C25-C39 nonexistent.

B. PHOSPHORYLASE KINASE Roach et af. (28, 29) were the first to demonstrate that phosphorylase kinase catalyses the phosphorylation of glycogen synthase, and this finding was con-

II.

465

MUSCLE GLYCOGEN SYNTHASE

firmed by others (23, 30, 31). The phosphorylation occurs at site-2 (23, 32) and the rate of phosphorylation is comparable to that of glycogen phosphorylase (23). The report of an additional phosphorylation site in the C-terminal region after prolonged incubation with high concentrations of phosphorylase kinase (32) can be explained by trace contamination with glycogen synthase kinase-5 (33). c . CALMODULIN-DEPENDENT “MULTIPROTEIN” KINASE The finding that purified preparations of glycogen synthase were contaminated with traces of a protein kinase that was stimulated by Ca2 and calmodulin (24, 34) led to the discovery of a Ca2 -calmoduIin-dependent glycogen synthase kinase in liver (35) and skeletal muscle (36).The rabbit skeletal-muscle enzyme has been purified -5000-fold and shows a major 58-kDa band and a minor 54kDa species when analyzed by SDS-PAGE. The native enzyme is a dodecamer with a molecular mass of 700 kDa, and the 12 subunits appear to be arranged as two hexagonal rings stacked one upon the other, as judged by electron microscopy (37). The enzyme phosphorylates site-2 and site-Ib, the initial rate of phosphorylation of site-2 being 5- to 10-fold faster than site-lb (37). The calmodulin-dependent glycogen synthase kinase has a broad substrate specificity in vitro, and is capable of phosphorylating a number of proteins at comparable rates to glycogen synthase. These include synapsin I, microtubuleassociated protein 2, and tyrosine hydroxylase (38,39). Glycogen phosphorylase is not a substrate (36, 37). Like many protein kinases the calmodulin-dependent glycogen synthase kinase can phosphorylate itself, and up to 5 mol phosphate/subunit are incorporated via the autophosphorylation reaction. Autophosphorylation does not affect activity measured in the presence of Ca2+ and calmodulin (37). However, recent work with the closely related brain enzyme (see below) has demonstrated that autophosphorylationcauses the protein kinase to become almost fully active in the absence of Ca2 and calmodulin (394. This may represent a mechanism for prolonging the Ca2+ signal. A synthetic peptide corresponding to the first 10 residues of glycogen synthase is an excellent substrate for the calmodulin-dependentglycogen synthase kinase, the phosphorylation occurring at N7 (40). If the arginine at N4 is substituted with leucine or alanine the peptide no longer serves as a substrate. Studies with other synthetic peptides have confirmed that the enzyme will only phosphorylate sequences of the type Arg-x-y-Ser-z at significant rates. However, in contrast to CAMP-dependent protein kinase, insertion of a second arginine residue at position x does not improve the kinetics of phosphorylation (40). A protein kinase with an identical substrate specificity is present in brain, where it has been termed synapsin I kinase-I1 (38) or calmodulin-dependent protein kinase-I1 (39). The brain enzyme is composed of two isoenzymes with subunit molecular masses of 50 kDa and 58-60 kDa whose proportions vary +

+

+

466

PHILIP COHEN

from brain region to brain region (41).The 58-60-kDa component from brain is closely related to the skeletal-muscle enzyme, as judged by one-dimensional peptide mapping of phosphopeptides and immunological criteria (38, 39), but does not seem to be identical (41a). The calmodulin-dependent glycogen synthase kinase has been purified from both rat and rabbit liver (42-44). The enzyme preparations show a proteinstaining doublet (57-55 kDa) on SDS-PAGE. However based on a sedimentation constant of 10.6 S and Stokes radius of 70 A, the native enzyme appears to have a molecular mass of =300 kDa (43). This suggests that it may be a hexamer, in contrast to the muscle enzyme which is a dodecamer. The liver enzyme also phosphorylates rabbit skeletal-muscle glycogen synthase (at site-2 and site-1b) (43), synapsin I, microtubule-associated protein 2, and tyrosine hydroxylase (45). The broad substrate specificity and widespread tissue distribution of this protein kinase suggests that it may mediate many of the actions of Ca2+ in vivo. Accordingly, it has been termed the calmodulin-dependent “multiprotein” kinase (38). D. GLYCOGEN SYNTHASE KINASE-3 Glycogen synthase kinase-3 (GSK-3) has been purified -50,000-fold to homogeneity from rabbit skeletal muscle. Its molecular mass estimated by SDSPAGE (51 kDa) is similar to that obtained by sedimentation equilibrium centrifugation of the native enzyme (47 kDa), demonstrating that GSK-3 is monomeric. However, it is eluted from gel filtration columns slightly earlier than serum albumin (66 kDa), indicating an asymmetric structure (18). GSK-3 phosphorylates the tryptic peptide comprising residues C28 to C39 specifically (13, 14). Following incubation of glycogen synthase with GSK-3 and MgATP, mono-, di-, and triphosphorylated forms of this tryptic peptide can be resolved, demonstrating that at least three seryl residues are phosphorylated (13).Based on the release of 32P-radioactivityduring automated Edman degradation of C28-C39, the residues phosphorylated appear to be C30, C34, and C38 (13);however, the order of phosphorylation is unknown. These serine residues are collectively referred to as sites-3. The type 11 regulatory subunit of CAMP-dependent protein kinase is a substrate for GSK-3 and two residues (Ser-44 and Ser-47) are phosphorylated (46). As discussed in Section VI, GSK-3 can also phosphorylate a protein termed inhibitor-2 on a specific threonyl residue (47). GSK-3 phosphorylates itself, and up to 4 phosphates/mol can be incorporated via autophosphorylation, without any apparent effect on activity ( 4 6 4 8 ) .However, many proteins that are phosphorylated by CAMP-dependent protein kinase are not touched by GSK-3 (49).Further properties of GSK-3 are reviewed in Ref. (49).GSK-3 has also been termed factor FAby Merlevede and co-workers (48, 50) for reasons discussed in Section VI.

11.

467


Ahmed et al. (51) partially purified a protein kinase from skeletal muscle whose activity was stimulated several-fold by heparin (AO.,=3pg/ml). Although it was originally suggested that this enzyme was distinguishable from other glycogen synthase kinases (51), more recent work suggests that it is GSK3 ( 5 1 ~ ) . The heparin-stimulated kinase phosphorylates the tryptic peptide C25-C39 specifically, and has Factor FA activity (see Section VI). Its apparent molecular mass estimated by gel filtration (70kDa) is identical to GSK3 (51). Highly purified GSK3 is not activated by heparin (33), but stimulation is lost during exposure to low ionic strength prior to chromatography on DEAE-cellulose. Loss of heparin stimulation is caused by a rise in activity in the absence of the glycosaminoglycan (C. Smythe, unpublished work from this laboratory). It therefore appears that the heparin-stimulated protein kinase may represent the “native” form of GSK3. However, no substances capable of substituting for heparin that might be of physiological importance have been identified so far. GSK-3 was partially purified from rabbit liver by DePaoli-Roach et al. (52). Their preparation predominantly phosphorylated the C-terminal cyanogen bromide peptide (CB-C) of rabbit skeletal-muscle glycogen synthase, as expected, but some incorporation of phosphate occurred in CB-N. Ramakrishna et al. (52a, 52b) also isolated a protein kinase from liver that phosphorylated site-2 in addition to sites-3, and this preparation was capable of phosphorylating other proteins that are not substrates for muscle GSK3, such as ATP-citrate lyase and acetylCoA carboxylase (52a, 52b). These results suggest that hepatic GSK-3 either has a broader specificity than its muscle counterpart or that the preparations are contaminated with another protein kinase.

E. GLYCOGEN SYNTHASE KINASE-4 Glycogen synthase kinase-4 (GSK-4) has been only partially purified from rabbit skeletal muscle, and its subunit composition is therefore unknown. The apparent molecular mass on gel filtration is 2 1 1 5 kDa. It phosphorylates glycogen synthase at site-2 and no other protein tested is phosphorylated at a significant rate (33). The substrate specificity of GSK-4 demonstrates that it is not a proteolytic fragment of phosphorylase kinase or the calmodulin-dependent multiprotein kinase that has lost its ability to be regulated by Ca2 -calmodulin (37). GSK-4 is identical to the enzyme termed PC,,, by Roach and co-workers (52) and to certain other glycogen synthase kinases that have been reported [reviewed in Ref. (33)].Further properties of GSK-4 are summarized in Ref. (49). No mechanisms for regulating the activity of GSK-4 have been identified. +

F. GLYCOGEN SYNTHASE KINASE-5 Glycogen synthase-5 (GSK-5) has been termed variously PC,., (52), casein . kinase-11, casein kinase-G, or casein kinase-TS [see discussion in Ref. ( 3 3 ) ] The

468

PHILIP COHEN

enzyme has been purified to homogeneity from skeletal muscle (54) and other tissues (55-57) and has an azPz structure in which the apparent molecular masses of the a- and P-subunits are 43 kDa and 26 kDa, respectively. The 43kDa component is the catalytic subunit (55, 58). GSK-5 has a number of distinctive properties including up to 40-fold activation by spermine at physiological (1 mM) concentrations of Mg2+, potent inhibition by heparin (Ki 9 0.05 pg/ml), and the ability to use GTP as a substrate almost as effectively as ATP (33, 55). GSK-5 phosphorylates glycogen synthase at residue C46, termed site-5 (14, 33). However, the enzyme has a broad substrate specificity, and physiological substrates include the type II regulatory subunit of CAMP-dependent protein kinase (46),troponin T (59, 60),and the P-subunit of protein synthesis initiation factor eIF-2 (60, 61). The enzyme can also phosphorylate its own @subunit without any effect on activity (53-55). GSK-5 phosphorylates seryl residues that are followed by a number of consecutive acidic residues (46), and this factor is critical for specific substrate recognition (18, 63, 64). G. CASEINKINASE-I Casein kinase-I (CK-I) has been purified = 100,000-fold to near homogeneity from skeletal muscle, and is a monomeric protein of molecular mass = 35 kDa (65), like CK-I from other mammalian sources (54, 66). It has also been termed glycogen synthase kinase-1 (67-70) or PC,,, (52, 71). Incubation of glycogen synthase with high concentrations of muscle or liver CK-I for 2-5 h results in the incorporation of 6 phosphates/subunit (65, 70, 71) and as many as 10 residues may become phosphorylated (65). In CB-N the seryl residues N3, N7, and NIO are major sites of phosphorylation and the threonyl residue N5, a minor site. The electrophoretic mobility of CB-N is slower after phosphorylation by CK-1 than after phosphorylation by CAMP-dependentprotein kinase. This implies that N7 cannot be the first residue in CB-1 phosphorylated by CK-I;the initial serine phosphorylated must therefore be either N3 or N10. The C-terminal cyanogen bromide peptide (CB-C) is phosphorylated by CK-I at a similar rate to CB-N, and at least five of the seven serines in the tryptic peptide C28-C53 are phosphorylated. These include the residues phosphorylated by GSK-3 and GSKJ. The exact seryl residues phosphorylated by CK-I are unknown, because phosphorylation renders the Arg-His bond between C39 and C40 completely resistant to trypsin (65). In contrast, the same bond is cleaved readily by trypsin if glycogen synthase is phosphorylated by either GSK-3 or GSK-5 (13, 33). This might suggest that C41 is phosphorylated by CK-I, because the failure of trypsin to cleave sequences of the type Arg-x-Ser(P) is well documented [e.g., see Refs. (72-74)]. Minor phosphorylation by CK-I also occurs in the tryptic peptide C98 to C123, mainly at seryl residues (65).

11.

469


Casein kinase-I has a very broad substrate specificity in v i m and can phosphorylate many proteins in addition to glycogen synthase [see Refs. (55, 75-77)]. However no mechanisms for regulating its activity have been identified. H.

CYCLICGUANOSINE MONOPHOSPHATE-DEPENDENT PROTEINKINASE

Cyclic guanosine monophosphate (cGMP) is present in extremely low concentrations in skeletal muscle (78, 79), and phosphorylation of glycogen synthase has only been examined using cGMP-dependent protein kinase from lung. The enzyme phosphorylates the tryptic peptides containing site-2, site-la, and site- 1b (26), the same peptides labeled by CAMP-dependent protein kinase. The order of phosphorylation is also site-la > site-2 > site-lb, although the difference in initial rate of phosphorylation between site-la and the other two sites is not as pronounced (26). These observations are not unexpected in view of the known similarity in substrate specificity between these two cyclic nucleotide-dependent protein kinases. However, the rate of phosphorylation of glycogen synthase by cGMP-dependent protein kinase is about 100-fold slower than that of CAMPdependent protein kinase (26, 79).

I. DIACYLGLYCEROL-DEPENDENT PROTEIN KINASE The activity of diacylglycerol (DG)-dependent protein kinase requires phosphatidylserine and supraphysiological concentrations of Ca2 , but in the presence of DG, the Ao,5 for Ca2+ is decreased over 1000-fold [reviewed in Ref. (SO)]. The activity of DG-dependent protein kinase is much lower in skeletal muscle than in other mammalian tissues (81, 82) and phosphorylation has only been examined using the rat brain enzyme (83). Glycogen synthase can be phosphorylated to >1 mol/subunit and the major tryptic peptides that become labeled are N5-N38 (containing site-2) and C85-C97 (containing site- la). The C-terminal peptide is phosphorylated at a slightly faster rate. Glycogen synthase is phosphorylated at a comparable rate to mixed histones, and as with most substrates, activity is stimulated -10-fold by Ca2+ and phospholipid (83). +

IV. Effect of Phosphorylation on the Activity of Skeletal-Muscle Glycogen Synthase

For many years it was believed that glycogen synthase existed in just two forms, a phosphorylated species dependent on G6P and a dephosphorylated form that was fully active in the absence of this effector. Subsequently, it was found that activation of the phosphorylated form by G6P could be antagonised by

470

PHILIP COHEN

metabolites such as ATP and Pi. The dephosphorylated form was also inhibited strongly by ATP, but this inhibition was reversed by very low concentrations of G6P (84).Thus, it appeared that glycogen synthase could exist in two forms that differed in sensitivity to G6P, ATP, and Pi. The most detailed study of the effects of phosphorylation on the kinetic properties of glycogen synthase (85, 86) was carried out before the multiplicity of glycogen synthase kinases was appreciated. In these experiments, glycogen synthase was phosphorylated in undefined sites by incubation with MgATP for varying periods of time, at an early stage of purification when the enzyme was still contaminated with protein kinase activities. Glycogen synthase was purified to homogeneity and its phosphate content and kinetic properties examined. It was found that the So,5 for UDP-Glc increased about 1000-fold over the phosphorylation range studied (0.27 to 3.5 mol/subunit). G6P attenuated the effect of phosphorylation on the So,5 for UDP-Glc although the Ao,5 for G6P also increased about 1000-fold over the same phosphorylation range and G6P-saturation curves became more sigmoidal. Phosphorylation was accompanied by a greater sensitivity to inhibition by substances such as ATP and Pi. However, even with highly phosphorylated enzyme, such inhibition could be counteracted effectively by G6P. An exhaustive kinetic analysis using glycogen synthase phosphorylated in defined sites by defined protein kinases, alone and in combination, has not been performed. Most studies have simply measured the “activity ratio” of the enzyme, defined as activity in the absence of G6P divided by activity in the presence of saturating G6P (usually measured at = 5 mM UDP-Glc). Cyclic AMP-dependent protein kinase phosphorylates site- la much faster than site-lb or site-2 (Section 111, A). Conversely, if glycogen synthase labeled in site- la, site- 1b, and site-2 is incubated with protein phosphatase-1 (Section VI), phosphate is removed sequentially from the three sites. The dephosphorylation of site-2 precedes site-la, and site-la precedes site-lb (26).These studies demonstrate that site-2 and site- la are both inactivating sites, although phosphorylation of site-2 depresses the activity ratio to a greater extent than site-la. In contrast, phosphorylation of site- 1b appears to have little or no effect on the activity ratio (26). There is general agreement that maximal phosphorylation of sites-3 by GSK-3 produces a greater decrease in the activity ratio than the phosphorylation of l b 2 (33, 52, 70). However, the effects of these phosphorylations site-la are additive, and greater decreases in the activity ratio are observed when all six sites are phosphorylated (87). Since the order of phosphorylation of C30, C34, and C38 is unknown, the relative contributions of these three sites to inactivation is unclear. Two laboratories have reported that phosphorylation of glycogen synthase by GSK-5 does not decrease the activity ratio (33, 54), whereas another group

+

+

11. MUSCLE GLYCOGEN SYNTHASE

47 1

reported that phosphorylation was accompanied by a small reduction in activity (56, 70). Phosphorylation of glycogen synthase to 4-6 mol/mol subunit by CK-I is accompanied by a decrease in activity ratio similar to that observed with GSK-3 and CAMP-dependent protein kinase combined (65, 67-71). This is consistent with the finding that CK-I phosphorylates serine residues in the region N3-N 10 and C30-C46 (Section 111, G).

V. Synergism between Glycogen Synthase Kinase-3 and Glycogen Synthase Kinase-5

Although the phosphorylation of site-5 by GSK-5 does not affect the activity ratio, the presence of phosphoserine at this position is critical for the activity of GSK-3. In this laboratory “dephosphorylated” preparations of glycogen synthase with activity ratios of 0 . 8 to 0.9 usually contain 0.5-0.6 mol phosphate/subunit, mostly located in the tryptic peptide (C40’ to C53) containing site-5 (88).This phosphate is resistant to the action of skeletal-muscle protein phosphatases (Section VI), but can be removed by incubation with potato acid phosphatase. This treatment abolishes phosphorylation of glycogen synthase by GSK-3, without affecting the rate of phosphorylation by CAMP-dependent protein kinase, phosphorylase kinase, the calmodulin-dependent multiprotein kinase, or GSK-4. Rephosphorylation at C46 by GSK-5 restores the ability of GSK-3 to phosphorylate the enzyme (89). The presence of = 0.5 mol phosphate/subunit in the tryptic peptide C40-C53 may explain why phosphorylation by GSK-3 usually reaches a plateau near 1.5 mol/subunit in v i m , rather than 3 mol/subunit. This view is supported by the work of DePaoli-Roach et al. (52), who found that phosphorylation by GSK-5 (without prior incubation with potato acid phosphatase) increased the amount of phosphate that could be incorporated by GSK-3 from 1.3 to > 2 mol/subunit. The effects of dephosphorylating and rephosphorylating the peptide C40-C53 on the phosphorylation of glycogen synthase by GSK-3 can be reproduced using the peptide C24-C124 (18) that can be isolated by brief chymotryptic attack of the native enzyme (Section 11). Since peptide C24-C124 is monomeric, this demonstrates that phosphorylation of C40-C53 is essential for phosphorylation of the same subunit by GSK-3 (18). These observations suggest that GSK-5 is a novel protein kinase, whose function is to form the recognition site for another protein kinase. Similar observations have been made for two other proteins. The type I1 regulatory subunit of CAMP-dependent protein kinase contains 1.5- 1.8 phosphates/subunit mostly located in Ser-74 and Ser-76, the sites phosphorylated by GSK-5. Dephosphorylation of these residues by incubation with potato acid phosphatase prevents

-

472

PHILIP COHEN

GSK-3 from phosphorylating Ser-44 and Ser-47. Rephosphorylation with GSK-5 restores the ability of GSK-3 to phosphorylate the protein (46). The amino acid sequence following Ser-76 (Glu-Asp-Glu-Glu-Asp) is almost identical to that following C46 (Fig. 1). DePaoli-Roach (90) reported that inhibitor-2 is a substrate for GSK-5, and that phosphorylation by this protein kinase potentiated phosphorylation by GSK-3 (see also Section VI). The residue phosphorylated by GSK-3 is threonine-72 (90a), and the residues phosphorylated by GSK-5 are serines 86, 120, and 121 (906). These findings suggest that the presence of a C-terminal phosphoserine residue is critical for substrate recognition by GSK-3. However, in the case of glycogen synthase a region C-terminal to residue C64 is also essential for phosphorylation (18).

VI. The Glycogen Synthase Phosphatases in Skeletal Muscle

Relatively few serine- and threonine-specificprotein phosphatases are present in the cytoplasm of mammalian cells (91, 92). Two of these enzymes, termed protein phosphatase- 1 (PP- 1) and protein phosphatase-2A (PP-2A) have broad substrate specificities and account for virtually all detectable glycogen synthase phosphatase and phosphorylase phosphatase activity in skeletal-muscle extracts. Thus the combined addition of inhibitor-2 (1-2), a specific inhibitor of PP-1, and antibody to PP-2A, inhibit glycogen synthase phosphatase and phosphorylase phosphatase activities in rabbit skeletal-muscleextracts by >95% (9.3). Furthermore, fractionation of the extracts by anion-exchange chromatography and gel filtration fails to detect any other protein phosphatase with significant activity toward glycogen synthase (94). A third protein phosphatase (PP-2C) capable of dephosphorylating glycogen synthase is present in skeletal-muscle extracts, but its contribution to the total activity (1 -2%) is negligible (94, 95). PP-1 and PP-2A dephosphorylate site-la, site-2, and sites-3 at comparable rates in virro (91). However, as discussed in Section IV, if glycogen synthase is phosphorylated in site-la + l b + 2, PP-1 dephosphorylates site-2 5- to 10-fold faster than site- la and = 100-fold faster than site-lb. The dephosphorylation of site-1a occurs more rapidly once site-2 is dephosphorylated, and dephosphorylation of site- 1b takes place at a significant rate only after both site-2 and site- 1a are dephosphorylated (26). Site-lb, and especially site-5 (91), are dephosphorylated very slowly by PP-1 and PP-2A, and no other protein phosphatases capable of acting on these sites at a significant rate have been detected in skeletal muscle. The extremely weak phosphatase activity toward site-5 may explain the high level of phosphorylation of the region C40-C53 in vivo (Section VII). The large

11.


473

number of acidic residues immediately C-terminal to C46 may act as a negative specificity determinant for protein phosphatases (91). The order of dephosphorylation of residues C30, C34, and C38 (sites-3), like their order of phosphorylation, is unknown. A. STRUCTURE OF PROTEIN PHOSPHATASE1 AND PROTEIN hOSPHATASE-2A The catalytic (C)-subunits of PP- I and PP-2A [termed C-I and C-I1 by Lee and co-workers (96)] have been purified to homogeneity by procedures that involve precipitation with 80% ethanol at room temperature at an early stage of purification (96, 97). If proteinase inhibitors are included, the C-subunit of PP-1 is recovered as a 37-kDa protein and the C-subunit of PP-2A as a 36-kDa species (97). However, despite their similar molecular mass and substrate specificities, peptide mapping studies have established that the two C-subunits are the products of distinct genes (97). The C-subunits do not exist as such in vivo, but are complexed with other proteins, removed, or denatured during treatment with 80% ethanol. Several of these high-molecular-mass forms have been purified to homogeneity from skeletal muscle and their subunit compositions elucidated. When skeletal-muscle extracts are centrifuged at 80,000 g, to pellet glycogen and its associated proteins, 50-60% of the PP-1 activity sediments with these glycogen-protein particles (12, 95). This is similar to the proportion of glycogen synthase bound to glycogen (22). The glycogen-bound form of PP-1, termed PP-I,, consists of the 37-kDa C-subunit complexed to a 103-kDa G-subunit, which is the glycogen-binding component (98). It is probable that much of the PP-1 activity that does not sediment with the glycogen-protein particles is also PP-I,, although this remains to be established. Protein phosphatase-1 can also be isolated in an inactive form, termed PP-I,, which is not associated with glycogen. It consists of the 37-kDa C-subunit complexed to 1-2 (99-102), whose molecular mass is 22.8 kDa ( 9 0 ~ )PP-1, . has also been termed the MgATP-dependent protein phosphatase, because preincubation with MgATP and another protein (factor FA) is required to generate catalytic activity (103). Factor FA has been purified to homogeneity and shown to be identical to GSK-3 (18, 48). Activation of PP-1, is triggered by the . mechanism of phosphorylation of a threonyl residue on 1-2 (18, 47, 9 0 ~ )The activation and deactivation of PP- 1, is discussed in greater detail elsewhere in this volume (Chapter 8). Protein phosphatase-2A is not associated with the glycogen-protein particles (95) and three forms of this enzyme can be resolved by chromatography on DEAE-cellulose, termed PP-2A0, PP-2A,, and PP-2A2 (91, 105). Each of these

474

PHILIP COHEN

species contain a 60-kDa A-subunit and a 36-kDa C-subunit. The A- and Csubunits of PP-2Ao, PP-2Al, and PP-2A2 are identical, as judged by peptide mapping, and the C-subunit is identical to the catalytic subunit of PP-2A isolated by treatment with 80% ethanol at room temperature. PP-2& contains an additional 54-kDa B’-subunit and PP-2A, a 55-kDa B-subunit. The B’- and Bsubunits display different peptide maps and therefore appear to be distinct gene products. PP-2A2 lacks the B’ and B-subunits, and appears to be derived from PP-2A0 and/or PP-2Al during purification, through dissociation and/or degradation of the B’- and/or B-subunits. Consequently PP-2Ao and PP-2A, may be the species that are present in vivo. PP-2& and PP-2Al have the subunit structures AB’C, and ABC,, respectively. The structure of PP-2A2 appears to be AC (105).

B . REGULATIONOF PROTEIN PHOSPHATASE1 AND PROTEIN PHOSPHATASE-2.4 Protein phosphatase- 1 is inhibited by nanomolar concentrations of a protein, termed inhibitor-1 (I-l), which functions as an inhibitor only if it is first phosphorylated by CAMP-dependent protein kinase (106, 107). The rate of phosphorylation of 1-1 in vitro is similar to that of glycogen synthase (108). The complete primary structure of 1-1 has been determined (109) and the site of phosphorylation is Thr-35. The concentration of 1-1 in muscle is = 1.8 pA4 (110), higher than that of PP-1, PP-l,, which is about 0.5 pA4 (97, 98). The state of phosphorylation of 1-1 in skeletal muscle is under hormonal control. Epinephrine increases the level of phosphorylation in vivo (111 ) or in the perfused hind limb (112, 113). The phosphorylation state of 1-1 in the perfused hind limb is exquisitely sensitive to P-adrenergic agonists, half-maximal effects being observed at 1 nM isoproterenol (113). A concentration of isoproterenol (0.5 nM), which produces a 40% increase in CAMP, causes a 2-fold rise in the phosphorylation state of 1-1 (from 15 to 30%). Both effects are prevented by nanomolar concentrations of insulin, added together with isoproterenol. However, at high levels of isoproterenol, where the level of phosphorylation is ==70%, or in the presence of the P-adrenergic antagonist, L-propranolol, where the level of phosphorylation is

The Enzymes: Control by Phosphorylation, Part A : General Features, Specific Enzymes

The enzymes. Vol. XVII, Control by phosphorylation: general features, specific enzymes (I)

The Enzymes, Vol XVIII: Control by Phosphorylation, Part B (Specific Enzymes), 3rd Edition

Enzymes

The Enzymes

Neurotransmitter Enzymes

Enzymes Everywhere

Artificial Enzymes

Understanding enzymes

The Enzymes, Volume 24: Protein Methyltransferases (The Enzymes)

The Enzymes, Vol XIV: Nucleic Acids, Part A, 3rd Edition

The Enzymes, Vol XIV: Nucleic Acids, Part A, 3rd Edition

Protein Prenylation, Volume 29: Part A (The Enzymes)

Proteolytic Enzymes.. A Practical Approach

Drug Metabolism Enzymes

Food Related Enzymes

Enzymes in Food Technology

Quinones and Quinone Enzymes

Enzymes in Food Technology

Enzymes for Carbohydrate Engineering

Enzymes in detergency

Enzymes in Detergency

Quinones and Quinone Enzymes

Textile Processing with Enzymes

Enzymes in Industry

Immobilized Enzymes and Cells

Enzymes in Lipid Modification

Enzymes of Molecular Biology

Immobilized Enzymes and Cells

The Enzymes: Control by Phosphorylation, Part A : General Features, Specific Enzymes

The enzymes. Vol. XVII, Control by phosphorylation: general features, specific enzymes (I)

The Enzymes, Vol XVIII: Control by Phosphorylation, Part B (Specific Enzymes), 3rd Edition

Enzymes

The Enzymes

Neurotransmitter Enzymes

Enzymes Everywhere

Artificial Enzymes

Understanding enzymes

The Enzymes, Volume 24: Protein Methyltransferases (The Enzymes)

The Enzymes, Vol XIV: Nucleic Acids, Part A, 3rd Edition

The Enzymes, Vol XIV: Nucleic Acids, Part A, 3rd Edition

Protein Prenylation, Volume 29: Part A (The Enzymes)

Proteolytic Enzymes.. A Practical Approach

Drug Metabolism Enzymes

Food Related Enzymes

Enzymes in Food Technology

Quinones and Quinone Enzymes

Enzymes in Food Technology

Enzymes for Carbohydrate Engineering

Enzymes in detergency

Enzymes in Detergency

Quinones and Quinone Enzymes

Textile Processing with Enzymes

Enzymes in Industry

Immobilized Enzymes and Cells

Enzymes in Lipid Modification

Enzymes of Molecular Biology

Immobilized Enzymes and Cells

Recommend Documents